Identification of Non-Linear Structures using Recurrent Neural Networks
DEFF Research Database (Denmark)
Kirkegaard, Poul Henning; Nielsen, Søren R. K.; Hansen, H. I.
Two different partially recurrent neural networks structured as Multi Layer Perceptrons (MLP) are investigated for time domain identification of a non-linear structure.......Two different partially recurrent neural networks structured as Multi Layer Perceptrons (MLP) are investigated for time domain identification of a non-linear structure....
Identification of Non-Linear Structures using Recurrent Neural Networks
DEFF Research Database (Denmark)
Kirkegaard, Poul Henning; Nielsen, Søren R. K.; Hansen, H. I.
1995-01-01
Two different partially recurrent neural networks structured as Multi Layer Perceptrons (MLP) are investigated for time domain identification of a non-linear structure.......Two different partially recurrent neural networks structured as Multi Layer Perceptrons (MLP) are investigated for time domain identification of a non-linear structure....
A recurrent neural network for solving bilevel linear programming problem.
He, Xing; Li, Chuandong; Huang, Tingwen; Li, Chaojie; Huang, Junjian
2014-04-01
In this brief, based on the method of penalty functions, a recurrent neural network (NN) modeled by means of a differential inclusion is proposed for solving the bilevel linear programming problem (BLPP). Compared with the existing NNs for BLPP, the model has the least number of state variables and simple structure. Using nonsmooth analysis, the theory of differential inclusions, and Lyapunov-like method, the equilibrium point sequence of the proposed NNs can approximately converge to an optimal solution of BLPP under certain conditions. Finally, the numerical simulations of a supply chain distribution model have shown excellent performance of the proposed recurrent NNs.
Sensitivity analysis of linear programming problem through a recurrent neural network
Das, Raja
2017-11-01
In this paper we study the recurrent neural network for solving linear programming problems. To achieve optimality in accuracy and also in computational effort, an algorithm is presented. We investigate the sensitivity analysis of linear programming problem through the neural network. A detailed example is also presented to demonstrate the performance of the recurrent neural network.
A novel recurrent neural network with finite-time convergence for linear programming.
Liu, Qingshan; Cao, Jinde; Chen, Guanrong
2010-11-01
In this letter, a novel recurrent neural network based on the gradient method is proposed for solving linear programming problems. Finite-time convergence of the proposed neural network is proved by using the Lyapunov method. Compared with the existing neural networks for linear programming, the proposed neural network is globally convergent to exact optimal solutions in finite time, which is remarkable and rare in the literature of neural networks for optimization. Some numerical examples are given to show the effectiveness and excellent performance of the new recurrent neural network.
International Nuclear Information System (INIS)
Liu, Xiaolan; Zhou, Mi
2016-01-01
In this paper, a one-layer recurrent network is proposed for solving a non-smooth convex optimization subject to linear inequality constraints. Compared with the existing neural networks for optimization, the proposed neural network is capable of solving more general convex optimization with linear inequality constraints. The convergence of the state variables of the proposed neural network to achieve solution optimality is guaranteed as long as the designed parameters in the model are larger than the derived lower bounds.
Xiao, Lin; Liao, Bolin; Li, Shuai; Chen, Ke
2018-02-01
In order to solve general time-varying linear matrix equations (LMEs) more efficiently, this paper proposes two nonlinear recurrent neural networks based on two nonlinear activation functions. According to Lyapunov theory, such two nonlinear recurrent neural networks are proved to be convergent within finite-time. Besides, by solving differential equation, the upper bounds of the finite convergence time are determined analytically. Compared with existing recurrent neural networks, the proposed two nonlinear recurrent neural networks have a better convergence property (i.e., the upper bound is lower), and thus the accurate solutions of general time-varying LMEs can be obtained with less time. At last, various different situations have been considered by setting different coefficient matrices of general time-varying LMEs and a great variety of computer simulations (including the application to robot manipulators) have been conducted to validate the better finite-time convergence of the proposed two nonlinear recurrent neural networks. Copyright © 2017 Elsevier Ltd. All rights reserved.
Directory of Open Access Journals (Sweden)
Daniel Durstewitz
2017-06-01
Full Text Available The computational and cognitive properties of neural systems are often thought to be implemented in terms of their (stochastic network dynamics. Hence, recovering the system dynamics from experimentally observed neuronal time series, like multiple single-unit recordings or neuroimaging data, is an important step toward understanding its computations. Ideally, one would not only seek a (lower-dimensional state space representation of the dynamics, but would wish to have access to its statistical properties and their generative equations for in-depth analysis. Recurrent neural networks (RNNs are a computationally powerful and dynamically universal formal framework which has been extensively studied from both the computational and the dynamical systems perspective. Here we develop a semi-analytical maximum-likelihood estimation scheme for piecewise-linear RNNs (PLRNNs within the statistical framework of state space models, which accounts for noise in both the underlying latent dynamics and the observation process. The Expectation-Maximization algorithm is used to infer the latent state distribution, through a global Laplace approximation, and the PLRNN parameters iteratively. After validating the procedure on toy examples, and using inference through particle filters for comparison, the approach is applied to multiple single-unit recordings from the rodent anterior cingulate cortex (ACC obtained during performance of a classical working memory task, delayed alternation. Models estimated from kernel-smoothed spike time data were able to capture the essential computational dynamics underlying task performance, including stimulus-selective delay activity. The estimated models were rarely multi-stable, however, but rather were tuned to exhibit slow dynamics in the vicinity of a bifurcation point. In summary, the present work advances a semi-analytical (thus reasonably fast maximum-likelihood estimation framework for PLRNNs that may enable to recover
Predicting non-linear dynamics by stable local learning in a recurrent spiking neural network.
Gilra, Aditya; Gerstner, Wulfram
2017-11-27
The brain needs to predict how the body reacts to motor commands, but how a network of spiking neurons can learn non-linear body dynamics using local, online and stable learning rules is unclear. Here, we present a supervised learning scheme for the feedforward and recurrent connections in a network of heterogeneous spiking neurons. The error in the output is fed back through fixed random connections with a negative gain, causing the network to follow the desired dynamics. The rule for Feedback-based Online Local Learning Of Weights (FOLLOW) is local in the sense that weight changes depend on the presynaptic activity and the error signal projected onto the postsynaptic neuron. We provide examples of learning linear, non-linear and chaotic dynamics, as well as the dynamics of a two-link arm. Under reasonable approximations, we show, using the Lyapunov method, that FOLLOW learning is uniformly stable, with the error going to zero asymptotically.
Chaotic diagonal recurrent neural network
International Nuclear Information System (INIS)
Wang Xing-Yuan; Zhang Yi
2012-01-01
We propose a novel neural network based on a diagonal recurrent neural network and chaos, and its structure and learning algorithm are designed. The multilayer feedforward neural network, diagonal recurrent neural network, and chaotic diagonal recurrent neural network are used to approach the cubic symmetry map. The simulation results show that the approximation capability of the chaotic diagonal recurrent neural network is better than the other two neural networks. (interdisciplinary physics and related areas of science and technology)
Dobbs, David E.
2013-01-01
A direct method is given for solving first-order linear recurrences with constant coefficients. The limiting value of that solution is studied as "n to infinity." This classroom note could serve as enrichment material for the typical introductory course on discrete mathematics that follows a calculus course.
Liu, Qingshan; Wang, Jun
2011-04-01
This paper presents a one-layer recurrent neural network for solving a class of constrained nonsmooth optimization problems with piecewise-linear objective functions. The proposed neural network is guaranteed to be globally convergent in finite time to the optimal solutions under a mild condition on a derived lower bound of a single gain parameter in the model. The number of neurons in the neural network is the same as the number of decision variables of the optimization problem. Compared with existing neural networks for optimization, the proposed neural network has a couple of salient features such as finite-time convergence and a low model complexity. Specific models for two important special cases, namely, linear programming and nonsmooth optimization, are also presented. In addition, applications to the shortest path problem and constrained least absolute deviation problem are discussed with simulation results to demonstrate the effectiveness and characteristics of the proposed neural network.
Ding, Lei; Xiao, Lin; Liao, Bolin; Lu, Rongbo; Peng, Hua
2017-01-01
To obtain the online solution of complex-valued systems of linear equation in complex domain with higher precision and higher convergence rate, a new neural network based on Zhang neural network (ZNN) is investigated in this paper. First, this new neural network for complex-valued systems of linear equation in complex domain is proposed and theoretically proved to be convergent within finite time. Then, the illustrative results show that the new neural network model has the higher precision and the higher convergence rate, as compared with the gradient neural network (GNN) model and the ZNN model. Finally, the application for controlling the robot using the proposed method for the complex-valued systems of linear equation is realized, and the simulation results verify the effectiveness and superiorness of the new neural network for the complex-valued systems of linear equation.
Predictions of SEP events by means of a linear filter and layer-recurrent neural network
Czech Academy of Sciences Publication Activity Database
Valach, F.; Revallo, M.; Hejda, Pavel; Bochníček, Josef
2011-01-01
Roč. 69, č. 9-10 (2011), s. 758-766 ISSN 0094-5765 R&D Projects: GA AV ČR(CZ) IAA300120608; GA MŠk OC09070 Grant - others:VEGA(SK) 2/0015/11; VEGA(SK) 2/0022/11 Institutional research plan: CEZ:AV0Z30120515 Keywords : coronal mass ejection * X-ray flare * solar energetic particles * artificial neural network Subject RIV: DE - Earth Magnetism, Geodesy, Geography Impact factor: 0.614, year: 2011
Liu, Tao; Huang, Jie
2017-04-17
This paper presents a discrete-time recurrent neural network approach to solving systems of linear equations with two features. First, the system of linear equations may not have a unique solution. Second, the system matrix is not known precisely, but a sequence of matrices that converges to the unknown system matrix exponentially is known. The problem is motivated from solving the output regulation problem for linear systems. Thus, an application of our main result leads to an online solution to the output regulation problem for linear systems.
Local Dynamics in Trained Recurrent Neural Networks.
Rivkind, Alexander; Barak, Omri
2017-06-23
Learning a task induces connectivity changes in neural circuits, thereby changing their dynamics. To elucidate task-related neural dynamics, we study trained recurrent neural networks. We develop a mean field theory for reservoir computing networks trained to have multiple fixed point attractors. Our main result is that the dynamics of the network's output in the vicinity of attractors is governed by a low-order linear ordinary differential equation. The stability of the resulting equation can be assessed, predicting training success or failure. As a consequence, networks of rectified linear units and of sigmoidal nonlinearities are shown to have diametrically different properties when it comes to learning attractors. Furthermore, a characteristic time constant, which remains finite at the edge of chaos, offers an explanation of the network's output robustness in the presence of variability of the internal neural dynamics. Finally, the proposed theory predicts state-dependent frequency selectivity in the network response.
Local Dynamics in Trained Recurrent Neural Networks
Rivkind, Alexander; Barak, Omri
2017-06-01
Learning a task induces connectivity changes in neural circuits, thereby changing their dynamics. To elucidate task-related neural dynamics, we study trained recurrent neural networks. We develop a mean field theory for reservoir computing networks trained to have multiple fixed point attractors. Our main result is that the dynamics of the network's output in the vicinity of attractors is governed by a low-order linear ordinary differential equation. The stability of the resulting equation can be assessed, predicting training success or failure. As a consequence, networks of rectified linear units and of sigmoidal nonlinearities are shown to have diametrically different properties when it comes to learning attractors. Furthermore, a characteristic time constant, which remains finite at the edge of chaos, offers an explanation of the network's output robustness in the presence of variability of the internal neural dynamics. Finally, the proposed theory predicts state-dependent frequency selectivity in the network response.
Recurrent Neural Network for Computing Outer Inverse.
Živković, Ivan S; Stanimirović, Predrag S; Wei, Yimin
2016-05-01
Two linear recurrent neural networks for generating outer inverses with prescribed range and null space are defined. Each of the proposed recurrent neural networks is based on the matrix-valued differential equation, a generalization of dynamic equations proposed earlier for the nonsingular matrix inversion, the Moore-Penrose inversion, as well as the Drazin inversion, under the condition of zero initial state. The application of the first approach is conditioned by the properties of the spectrum of a certain matrix; the second approach eliminates this drawback, though at the cost of increasing the number of matrix operations. The cases corresponding to the most common generalized inverses are defined. The conditions that ensure stability of the proposed neural network are presented. Illustrative examples present the results of numerical simulations.
Deep Gate Recurrent Neural Network
2016-11-22
and Fred Cummins. Learning to forget: Continual prediction with lstm . Neural computation, 12(10):2451–2471, 2000. Alex Graves. Generating sequences...DSGU) and Simple Gated Unit (SGU), which are structures for learning long-term dependencies. Compared to traditional Long Short-Term Memory ( LSTM ) and...Gated Recurrent Unit (GRU), both structures require fewer parameters and less computation time in sequence classification tasks. Unlike GRU and LSTM
Ocean wave forecasting using recurrent neural networks
Digital Repository Service at National Institute of Oceanography (India)
Mandal, S.; Prabaharan, N.
, merchant vessel routing, nearshore construction, etc. more efficiently and safely. This paper describes an artificial neural network, namely recurrent neural network with rprop update algorithm and is applied for wave forecasting. Measured ocean waves off...
Adaptive Filtering Using Recurrent Neural Networks
Parlos, Alexander G.; Menon, Sunil K.; Atiya, Amir F.
2005-01-01
A method for adaptive (or, optionally, nonadaptive) filtering has been developed for estimating the states of complex process systems (e.g., chemical plants, factories, or manufacturing processes at some level of abstraction) from time series of measurements of system inputs and outputs. The method is based partly on the fundamental principles of the Kalman filter and partly on the use of recurrent neural networks. The standard Kalman filter involves an assumption of linearity of the mathematical model used to describe a process system. The extended Kalman filter accommodates a nonlinear process model but still requires linearization about the state estimate. Both the standard and extended Kalman filters involve the often unrealistic assumption that process and measurement noise are zero-mean, Gaussian, and white. In contrast, the present method does not involve any assumptions of linearity of process models or of the nature of process noise; on the contrary, few (if any) assumptions are made about process models, noise models, or the parameters of such models. In this regard, the method can be characterized as one of nonlinear, nonparametric filtering. The method exploits the unique ability of neural networks to approximate nonlinear functions. In a given case, the process model is limited mainly by limitations of the approximation ability of the neural networks chosen for that case. Moreover, despite the lack of assumptions regarding process noise, the method yields minimum- variance filters. In that they do not require statistical models of noise, the neural- network-based state filters of this method are comparable to conventional nonlinear least-squares estimators.
Interpretation of Recurrent Neural Networks
DEFF Research Database (Denmark)
Pedersen, Morten With; Larsen, Jan
1997-01-01
This paper addresses techniques for interpretation and characterization of trained recurrent nets for time series problems. In particular, we focus on assessment of effective memory and suggest an operational definition of memory. Further we discuss the evaluation of learning curves. Various nume...
Families of Linear Recurrences for Catalan Numbers
Gauthier, N.
2011-01-01
Four different families of linear recurrences are derived for Catalan numbers. The derivations rest on John Riordan's 1973 generalization of Catalan numbers to a set of polynomials. Elementary differential and integral calculus techniques are used and the results should be of interest to teachers and students of introductory courses in calculus…
Analysis of surface ozone using a recurrent neural network.
Biancofiore, Fabio; Verdecchia, Marco; Di Carlo, Piero; Tomassetti, Barbara; Aruffo, Eleonora; Busilacchio, Marcella; Bianco, Sebastiano; Di Tommaso, Sinibaldo; Colangeli, Carlo
2015-05-01
Hourly concentrations of ozone (O₃) and nitrogen dioxide (NO₂) have been measured for 16 years, from 1998 to 2013, in a seaside town in central Italy. The seasonal trends of O₃ and NO₂ recorded in this period have been studied. Furthermore, we used the data collected during one year (2005), to define the characteristics of a multiple linear regression model and a neural network model. Both models are used to model the hourly O₃ concentration, using, two scenarios: 1) in the first as inputs, only meteorological parameters and 2) in the second adding photochemical parameters at those of the first scenario. In order to evaluate the performance of the model four statistical criteria are used: correlation coefficient, fractional bias, normalized mean squared error and a factor of two. All the criteria show that the neural network gives better results, compared to the regression model, in all the model scenarios. Predictions of O₃ have been carried out by many authors using a feed forward neural architecture. In this paper we show that a recurrent architecture significantly improves the performances of neural predictors. Using only the meteorological parameters as input, the recurrent architecture shows performance better than the multiple linear regression model that uses meteorological and photochemical data as input, making the neural network model with recurrent architecture a more useful tool in areas where only weather measurements are available. Finally, we used the neural network model to forecast the O₃ hourly concentrations 1, 3, 6, 12, 24 and 48 h ahead. The performances of the model in predicting O₃ levels are discussed. Emphasis is given to the possibility of using the neural network model in operational ways in areas where only meteorological data are available, in order to predict O₃ also in sites where it has not been measured yet. Copyright © 2015 Elsevier B.V. All rights reserved.
Supervised Sequence Labelling with Recurrent Neural Networks
Graves, Alex
2012-01-01
Supervised sequence labelling is a vital area of machine learning, encompassing tasks such as speech, handwriting and gesture recognition, protein secondary structure prediction and part-of-speech tagging. Recurrent neural networks are powerful sequence learning tools—robust to input noise and distortion, able to exploit long-range contextual information—that would seem ideally suited to such problems. However their role in large-scale sequence labelling systems has so far been auxiliary. The goal of this book is a complete framework for classifying and transcribing sequential data with recurrent neural networks only. Three main innovations are introduced in order to realise this goal. Firstly, the connectionist temporal classification output layer allows the framework to be trained with unsegmented target sequences, such as phoneme-level speech transcriptions; this is in contrast to previous connectionist approaches, which were dependent on error-prone prior segmentation. Secondly, multidimensional...
Embedding recurrent neural networks into predator-prey models.
Moreau, Yves; Louiès, Stephane; Vandewalle, Joos; Brenig, Leon
1999-03-01
We study changes of coordinates that allow the embedding of ordinary differential equations describing continuous-time recurrent neural networks into differential equations describing predator-prey models-also called Lotka-Volterra systems. We transform the equations for the neural network first into quasi-monomial form (Brenig, L. (1988). Complete factorization and analytic solutions of generalized Lotka-Volterra equations. Physics Letters A, 133(7-8), 378-382), where we express the vector field of the dynamical system as a linear combination of products of powers of the variables. In practice, this transformation is possible only if the activation function is the hyperbolic tangent or the logistic sigmoid. From this quasi-monomial form, we can directly transform the system further into Lotka-Volterra equations. The resulting Lotka-Volterra system is of higher dimension than the original system, but the behavior of its first variables is equivalent to the behavior of the original neural network. We expect that this transformation will permit the application of existing techniques for the analysis of Lotka-Volterra systems to recurrent neural networks. Furthermore, our results show that Lotka-Volterra systems are universal approximators of dynamical systems, just as are continuous-time neural networks.
Collaborative Recurrent Neural Networks forDynamic Recommender Systems
2016-11-22
JMLR: Workshop and Conference Proceedings 63:366–381, 2016 ACML 2016 Collaborative Recurrent Neural Networks for Dynamic Recommender Systems Young...an unprece- dented scale. Although such activity logs are abundantly available, most approaches to recommender systems are based on the rating...Recurrent Neural Network, Recommender System , Neural Language Model, Collaborative Filtering 1. Introduction As ever larger parts of the population
Estimating Ads’ Click through Rate with Recurrent Neural Network
Directory of Open Access Journals (Sweden)
Chen Qiao-Hong
2016-01-01
Full Text Available With the development of the Internet, online advertising spreads across every corner of the world, the ads' click through rate (CTR estimation is an important method to improve the online advertising revenue. Compared with the linear model, the nonlinear models can study much more complex relationships between a large number of nonlinear characteristics, so as to improve the accuracy of the estimation of the ads’ CTR. The recurrent neural network (RNN based on Long-Short Term Memory (LSTM is an improved model of the feedback neural network with ring structure. The model overcomes the problem of the gradient of the general RNN. Experiments show that the RNN based on LSTM exceeds the linear models, and it can effectively improve the estimation effect of the ads’ click through rate.
Neural Networks for Non-linear Control
DEFF Research Database (Denmark)
Sørensen, O.
1994-01-01
This paper describes how a neural network, structured as a Multi Layer Perceptron, is trained to predict, simulate and control a non-linear process.......This paper describes how a neural network, structured as a Multi Layer Perceptron, is trained to predict, simulate and control a non-linear process....
Analysis of Recurrent Analog Neural Networks
Directory of Open Access Journals (Sweden)
Z. Raida
1998-06-01
Full Text Available In this paper, an original rigorous analysis of recurrent analog neural networks, which are built from opamp neurons, is presented. The analysis, which comes from the approximate model of the operational amplifier, reveals causes of possible non-stable states and enables to determine convergence properties of the network. Results of the analysis are discussed in order to enable development of original robust and fast analog networks. In the analysis, the special attention is turned to the examination of the influence of real circuit elements and of the statistical parameters of processed signals to the parameters of the network.
Optimizing Markovian modeling of chaotic systems with recurrent neural networks
International Nuclear Information System (INIS)
Cechin, Adelmo L.; Pechmann, Denise R.; Oliveira, Luiz P.L. de
2008-01-01
In this paper, we propose a methodology for optimizing the modeling of an one-dimensional chaotic time series with a Markov Chain. The model is extracted from a recurrent neural network trained for the attractor reconstructed from the data set. Each state of the obtained Markov Chain is a region of the reconstructed state space where the dynamics is approximated by a specific piecewise linear map, obtained from the network. The Markov Chain represents the dynamics of the time series in its statistical essence. An application to a time series resulted from Lorenz system is included
Fine-tuning and the stability of recurrent neural networks.
Directory of Open Access Journals (Sweden)
David MacNeil
Full Text Available A central criticism of standard theoretical approaches to constructing stable, recurrent model networks is that the synaptic connection weights need to be finely-tuned. This criticism is severe because proposed rules for learning these weights have been shown to have various limitations to their biological plausibility. Hence it is unlikely that such rules are used to continuously fine-tune the network in vivo. We describe a learning rule that is able to tune synaptic weights in a biologically plausible manner. We demonstrate and test this rule in the context of the oculomotor integrator, showing that only known neural signals are needed to tune the weights. We demonstrate that the rule appropriately accounts for a wide variety of experimental results, and is robust under several kinds of perturbation. Furthermore, we show that the rule is able to achieve stability as good as or better than that provided by the linearly optimal weights often used in recurrent models of the integrator. Finally, we discuss how this rule can be generalized to tune a wide variety of recurrent attractor networks, such as those found in head direction and path integration systems, suggesting that it may be used to tune a wide variety of stable neural systems.
Deep Recurrent Convolutional Neural Network: Improving Performance For Speech Recognition
Zhang, Zewang; Sun, Zheng; Liu, Jiaqi; Chen, Jingwen; Huo, Zhao; Zhang, Xiao
2016-01-01
A deep learning approach has been widely applied in sequence modeling problems. In terms of automatic speech recognition (ASR), its performance has significantly been improved by increasing large speech corpus and deeper neural network. Especially, recurrent neural network and deep convolutional neural network have been applied in ASR successfully. Given the arising problem of training speed, we build a novel deep recurrent convolutional network for acoustic modeling and then apply deep resid...
Precipitation Nowcast using Deep Recurrent Neural Network
Akbari Asanjan, A.; Yang, T.; Gao, X.; Hsu, K. L.; Sorooshian, S.
2016-12-01
An accurate precipitation nowcast (0-6 hours) with a fine temporal and spatial resolution has always been an important prerequisite for flood warning, streamflow prediction and risk management. Most of the popular approaches used for forecasting precipitation can be categorized into two groups. One type of precipitation forecast relies on numerical modeling of the physical dynamics of atmosphere and another is based on empirical and statistical regression models derived by local hydrologists or meteorologists. Given the recent advances in artificial intelligence, in this study a powerful Deep Recurrent Neural Network, termed as Long Short-Term Memory (LSTM) model, is creatively used to extract the patterns and forecast the spatial and temporal variability of Cloud Top Brightness Temperature (CTBT) observed from GOES satellite. Then, a 0-6 hours precipitation nowcast is produced using a Precipitation Estimation from Remote Sensing Information using Artificial Neural Network (PERSIANN) algorithm, in which the CTBT nowcast is used as the PERSIANN algorithm's raw inputs. Two case studies over the continental U.S. have been conducted that demonstrate the improvement of proposed approach as compared to a classical Feed Forward Neural Network and a couple simple regression models. The advantages and disadvantages of the proposed method are summarized with regard to its capability of pattern recognition through time, handling of vanishing gradient during model learning, and working with sparse data. The studies show that the LSTM model performs better than other methods, and it is able to learn the temporal evolution of the precipitation events through over 1000 time lags. The uniqueness of PERSIANN's algorithm enables an alternative precipitation nowcast approach as demonstrated in this study, in which the CTBT prediction is produced and used as the inputs for generating precipitation nowcast.
Time series prediction with simple recurrent neural networks ...
African Journals Online (AJOL)
A hybrid of the two called Elman-Jordan (or Multi-recurrent) neural network is also being used. In this study, we evaluated the performance of these neural networks on three established bench mark time series prediction problems. Results from the experiments showed that Jordan neural network performed significantly ...
Directory of Open Access Journals (Sweden)
Xing Yin
2011-01-01
uncertain periodic switched recurrent neural networks with time-varying delays. When uncertain discrete-time recurrent neural network is a periodic system, it is expressed as switched neural network for the finite switching state. Based on the switched quadratic Lyapunov functional approach (SQLF and free-weighting matrix approach (FWM, some linear matrix inequality criteria are found to guarantee the delay-dependent asymptotical stability of these systems. Two examples illustrate the exactness of the proposed criteria.
Deep Recurrent Neural Networks for Supernovae Classification
Charnock, Tom; Moss, Adam
2017-03-01
We apply deep recurrent neural networks, which are capable of learning complex sequential information, to classify supernovae (code available at https://github.com/adammoss/supernovae). The observational time and filter fluxes are used as inputs to the network, but since the inputs are agnostic, additional data such as host galaxy information can also be included. Using the Supernovae Photometric Classification Challenge (SPCC) data, we find that deep networks are capable of learning about light curves, however the performance of the network is highly sensitive to the amount of training data. For a training size of 50% of the representational SPCC data set (around 104 supernovae) we obtain a type-Ia versus non-type-Ia classification accuracy of 94.7%, an area under the Receiver Operating Characteristic curve AUC of 0.986 and an SPCC figure-of-merit F 1 = 0.64. When using only the data for the early-epoch challenge defined by the SPCC, we achieve a classification accuracy of 93.1%, AUC of 0.977, and F 1 = 0.58, results almost as good as with the whole light curve. By employing bidirectional neural networks, we can acquire impressive classification results between supernovae types I, II and III at an accuracy of 90.4% and AUC of 0.974. We also apply a pre-trained model to obtain classification probabilities as a function of time and show that it can give early indications of supernovae type. Our method is competitive with existing algorithms and has applications for future large-scale photometric surveys.
Bayesian Recurrent Neural Network for Language Modeling.
Chien, Jen-Tzung; Ku, Yuan-Chu
2016-02-01
A language model (LM) is calculated as the probability of a word sequence that provides the solution to word prediction for a variety of information systems. A recurrent neural network (RNN) is powerful to learn the large-span dynamics of a word sequence in the continuous space. However, the training of the RNN-LM is an ill-posed problem because of too many parameters from a large dictionary size and a high-dimensional hidden layer. This paper presents a Bayesian approach to regularize the RNN-LM and apply it for continuous speech recognition. We aim to penalize the too complicated RNN-LM by compensating for the uncertainty of the estimated model parameters, which is represented by a Gaussian prior. The objective function in a Bayesian classification network is formed as the regularized cross-entropy error function. The regularized model is constructed not only by calculating the regularized parameters according to the maximum a posteriori criterion but also by estimating the Gaussian hyperparameter by maximizing the marginal likelihood. A rapid approximation to a Hessian matrix is developed to implement the Bayesian RNN-LM (BRNN-LM) by selecting a small set of salient outer-products. The proposed BRNN-LM achieves a sparser model than the RNN-LM. Experiments on different corpora show the robustness of system performance by applying the rapid BRNN-LM under different conditions.
Character recognition from trajectory by recurrent spiking neural networks.
Jiangrong Shen; Kang Lin; Yueming Wang; Gang Pan
2017-07-01
Spiking neural networks are biologically plausible and power-efficient on neuromorphic hardware, while recurrent neural networks have been proven to be efficient on time series data. However, how to use the recurrent property to improve the performance of spiking neural networks is still a problem. This paper proposes a recurrent spiking neural network for character recognition using trajectories. In the network, a new encoding method is designed, in which varying time ranges of input streams are used in different recurrent layers. This is able to improve the generalization ability of our model compared with general encoding methods. The experiments are conducted on four groups of the character data set from University of Edinburgh. The results show that our method can achieve a higher average recognition accuracy than existing methods.
A recurrent neural network for adaptive beamforming and array correction.
Che, Hangjun; Li, Chuandong; He, Xing; Huang, Tingwen
2016-08-01
In this paper, a recurrent neural network (RNN) is proposed for solving adaptive beamforming problem. In order to minimize sidelobe interference, the problem is described as a convex optimization problem based on linear array model. RNN is designed to optimize system's weight values in the feasible region which is derived from arrays' state and plane wave's information. The new algorithm is proven to be stable and converge to optimal solution in the sense of Lyapunov. So as to verify new algorithm's performance, we apply it to beamforming under array mismatch situation. Comparing with other optimization algorithms, simulations suggest that RNN has strong ability to search for exact solutions under the condition of large scale constraints. Copyright © 2016 Elsevier Ltd. All rights reserved.
Global robust exponential stability analysis for interval recurrent neural networks
International Nuclear Information System (INIS)
Xu Shengyuan; Lam, James; Ho, Daniel W.C.; Zou Yun
2004-01-01
This Letter investigates the problem of robust global exponential stability analysis for interval recurrent neural networks (RNNs) via the linear matrix inequality (LMI) approach. The values of the time-invariant uncertain parameters are assumed to be bounded within given compact sets. An improved condition for the existence of a unique equilibrium point and its global exponential stability of RNNs with known parameters is proposed. Based on this, a sufficient condition for the global robust exponential stability for interval RNNs is obtained. Both of the conditions are expressed in terms of LMIs, which can be checked easily by various recently developed convex optimization algorithms. Examples are provided to demonstrate the reduced conservatism of the proposed exponential stability condition
Representation of linguistic form and function in recurrent neural networks
Kadar, Akos; Chrupala, Grzegorz; Alishahi, Afra
2017-01-01
We present novel methods for analyzing the activation patterns of recurrent neural networks from a linguistic point of view and explore the types of linguistic structure they learn. As a case study, we use a standard standalone language model, and a multi-task gated recurrent network architecture
International Nuclear Information System (INIS)
Cui Baotong; Lou Xuyang
2009-01-01
In this paper, a new method to synchronize two identical chaotic recurrent neural networks is proposed. Using the drive-response concept, a nonlinear feedback control law is derived to achieve the state synchronization of the two identical chaotic neural networks. Furthermore, based on the Lyapunov method, a delay independent sufficient synchronization condition in terms of linear matrix inequality (LMI) is obtained. A numerical example with graphical illustrations is given to illuminate the presented synchronization scheme
On a class of fourth order linear recurrence equations
Directory of Open Access Journals (Sweden)
Sui-Sun Cheng
1984-01-01
Full Text Available This paper is concerned with sequences that satisfy a class of fourth order linear recurrence equations. Basic properties of such sequences are derived. In addition, we discuss the oscillatory and nonoscillatory behavior of such sequences.
Recurrent Neural Network Applications for Astronomical Time Series
Protopapas, Pavlos
2017-06-01
The benefits of good predictive models in astronomy lie in early event prediction systems and effective resource allocation. Current time series methods applicable to regular time series have not evolved to generalize for irregular time series. In this talk, I will describe two Recurrent Neural Network methods, Long Short-Term Memory (LSTM) and Echo State Networks (ESNs) for predicting irregular time series. Feature engineering along with a non-linear modeling proved to be an effective predictor. For noisy time series, the prediction is improved by training the network on error realizations using the error estimates from astronomical light curves. In addition to this, we propose a new neural network architecture to remove correlation from the residuals in order to improve prediction and compensate for the noisy data. Finally, I show how to set hyperparameters for a stable and performant solution correctly. In this work, we circumvent this obstacle by optimizing ESN hyperparameters using Bayesian optimization with Gaussian Process priors. This automates the tuning procedure, enabling users to employ the power of RNN without needing an in-depth understanding of the tuning procedure.
DEFF Research Database (Denmark)
Chon, K H; Hoyer, D; Armoundas, A A
1999-01-01
In this study, we introduce a new approach for estimating linear and nonlinear stochastic autoregressive moving average (ARMA) model parameters, given a corrupt signal, using artificial recurrent neural networks. This new approach is a two-step approach in which the parameters of the deterministic...... part of the stochastic ARMA model are first estimated via a three-layer artificial neural network (deterministic estimation step) and then reestimated using the prediction error as one of the inputs to the artificial neural networks in an iterative algorithm (stochastic estimation step). The prediction...... error is obtained by subtracting the corrupt signal of the estimated ARMA model obtained via the deterministic estimation step from the system output response. We present computer simulation examples to show the efficacy of the proposed stochastic recurrent neural network approach in obtaining accurate...
Noise-enhanced categorization in a recurrently reconnected neural network
International Nuclear Information System (INIS)
Monterola, Christopher; Zapotocky, Martin
2005-01-01
We investigate the interplay of recurrence and noise in neural networks trained to categorize spatial patterns of neural activity. We develop the following procedure to demonstrate how, in the presence of noise, the introduction of recurrence permits to significantly extend and homogenize the operating range of a feed-forward neural network. We first train a two-level perceptron in the absence of noise. Following training, we identify the input and output units of the feed-forward network, and thus convert it into a two-layer recurrent network. We show that the performance of the reconnected network has features reminiscent of nondynamic stochastic resonance: the addition of noise enables the network to correctly categorize stimuli of subthreshold strength, with optimal noise magnitude significantly exceeding the stimulus strength. We characterize the dynamics leading to this effect and contrast it to the behavior of a more simple associative memory network in which noise-mediated categorization fails
Noise-enhanced categorization in a recurrently reconnected neural network
Monterola, Christopher; Zapotocky, Martin
2005-03-01
We investigate the interplay of recurrence and noise in neural networks trained to categorize spatial patterns of neural activity. We develop the following procedure to demonstrate how, in the presence of noise, the introduction of recurrence permits to significantly extend and homogenize the operating range of a feed-forward neural network. We first train a two-level perceptron in the absence of noise. Following training, we identify the input and output units of the feed-forward network, and thus convert it into a two-layer recurrent network. We show that the performance of the reconnected network has features reminiscent of nondynamic stochastic resonance: the addition of noise enables the network to correctly categorize stimuli of subthreshold strength, with optimal noise magnitude significantly exceeding the stimulus strength. We characterize the dynamics leading to this effect and contrast it to the behavior of a more simple associative memory network in which noise-mediated categorization fails.
International Nuclear Information System (INIS)
Wang Linshan; Zhang Zhe; Wang Yangfan
2008-01-01
Some criteria for the global stochastic exponential stability of the delayed reaction-diffusion recurrent neural networks with Markovian jumping parameters are presented. The jumping parameters considered here are generated from a continuous-time discrete-state homogeneous Markov process, which are governed by a Markov process with discrete and finite state space. By employing a new Lyapunov-Krasovskii functional, a linear matrix inequality (LMI) approach is developed to establish some easy-to-test criteria of global exponential stability in the mean square for the stochastic neural networks. The criteria are computationally efficient, since they are in the forms of some linear matrix inequalities
Statistical downscaling of precipitation using long short-term memory recurrent neural networks
Misra, Saptarshi; Sarkar, Sudeshna; Mitra, Pabitra
2017-11-01
Hydrological impacts of global climate change on regional scale are generally assessed by downscaling large-scale climatic variables, simulated by General Circulation Models (GCMs), to regional, small-scale hydrometeorological variables like precipitation, temperature, etc. In this study, we propose a new statistical downscaling model based on Recurrent Neural Network with Long Short-Term Memory which captures the spatio-temporal dependencies in local rainfall. The previous studies have used several other methods such as linear regression, quantile regression, kernel regression, beta regression, and artificial neural networks. Deep neural networks and recurrent neural networks have been shown to be highly promising in modeling complex and highly non-linear relationships between input and output variables in different domains and hence we investigated their performance in the task of statistical downscaling. We have tested this model on two datasets—one on precipitation in Mahanadi basin in India and the second on precipitation in Campbell River basin in Canada. Our autoencoder coupled long short-term memory recurrent neural network model performs the best compared to other existing methods on both the datasets with respect to temporal cross-correlation, mean squared error, and capturing the extremes.
Optimization of recurrent neural networks for time series modeling
DEFF Research Database (Denmark)
Pedersen, Morten With
1997-01-01
The present thesis is about optimization of recurrent neural networks applied to time series modeling. In particular is considered fully recurrent networks working from only a single external input, one layer of nonlinear hidden units and a li near output unit applied to prediction of discrete time...... series. The overall objective s are to improve training by application of second-order methods and to improve generalization ability by architecture optimization accomplished by pruning. The major topics covered in the thesis are: 1. The problem of training recurrent networks is analyzed from a numerical...... of solution obtained as well as computation time required. 3. A theoretical definition of the generalization error for recurrent networks is provided. This definition justifies a commonly adopted approach for estimating generalization ability. 4. The viability of pruning recurrent networks by the Optimal...
Energy Complexity of Recurrent Neural Networks
Czech Academy of Sciences Publication Activity Database
Šíma, Jiří
2014-01-01
Roč. 26, č. 5 (2014), s. 953-973 ISSN 0899-7667 R&D Projects: GA ČR GAP202/10/1333 Institutional support: RVO:67985807 Keywords : neural network * finite automaton * energy complexity * optimal size Subject RIV: IN - Informatics, Computer Science Impact factor: 2.207, year: 2014
Neural Machine Translation with Recurrent Attention Modeling
Yang, Zichao; Hu, Zhiting; Deng, Yuntian; Dyer, Chris; Smola, Alex
2016-01-01
Knowing which words have been attended to in previous time steps while generating a translation is a rich source of information for predicting what words will be attended to in the future. We improve upon the attention model of Bahdanau et al. (2014) by explicitly modeling the relationship between previous and subsequent attention levels for each word using one recurrent network per input word. This architecture easily captures informative features, such as fertility and regularities in relat...
Bach in 2014: Music Composition with Recurrent Neural Network
Liu, I-Ting; Ramakrishnan, Bhiksha
2014-01-01
We propose a framework for computer music composition that uses resilient propagation (RProp) and long short term memory (LSTM) recurrent neural network. In this paper, we show that LSTM network learns the structure and characteristics of music pieces properly by demonstrating its ability to recreate music. We also show that predicting existing music using RProp outperforms Back propagation through time (BPTT).
Probing the basins of attraction of a recurrent neural network
Heerema, M.; van Leeuwen, W.A.
2000-01-01
Analytical expressions for the weights $w_{ij}(b)$ of the connections of a recurrent neural network are found by taking explicitly into account basins of attraction, the size of which is characterized by a basin parameter $b$. It is shown that a network with $b \
Bayesian model ensembling using meta-trained recurrent neural networks
Ambrogioni, L.; Berezutskaya, Y.; Gü ç lü , U.; Borne, E.W.P. van den; Gü ç lü tü rk, Y.; Gerven, M.A.J. van; Maris, E.G.G.
2017-01-01
In this paper we demonstrate that a recurrent neural network meta-trained on an ensemble of arbitrary classification tasks can be used as an approximation of the Bayes optimal classifier. This result is obtained by relying on the framework of e-free approximate Bayesian inference, where the Bayesian
Railway track circuit fault diagnosis using recurrent neural networks
de Bruin, T.D.; Verbert, K.A.J.; Babuska, R.
2017-01-01
Timely detection and identification of faults in railway track circuits are crucial for the safety and availability of railway networks. In this paper, the use of the long-short-term memory (LSTM) recurrent neural network is proposed to accomplish these tasks based on the commonly available
A recurrent neural network with ever changing synapses
Heerema, M.; van Leeuwen, W.A.
2000-01-01
A recurrent neural network with noisy input is studied analytically, on the basis of a Discrete Time Master Equation. The latter is derived from a biologically realizable learning rule for the weights of the connections. In a numerical study it is found that the fixed points of the dynamics of the
Active Control of Sound based on Diagonal Recurrent Neural Network
Jayawardhana, Bayu; Xie, Lihua; Yuan, Shuqing
2002-01-01
Recurrent neural network has been known for its dynamic mapping and better suited for nonlinear dynamical system. Nonlinear controller may be needed in cases where the actuators exhibit the nonlinear characteristics, or in cases when the structure to be controlled exhibits nonlinear behavior. The
Convolutional over Recurrent Encoder for Neural Machine Translation
Directory of Open Access Journals (Sweden)
Dakwale Praveen
2017-06-01
Full Text Available Neural machine translation is a recently proposed approach which has shown competitive results to traditional MT approaches. Standard neural MT is an end-to-end neural network where the source sentence is encoded by a recurrent neural network (RNN called encoder and the target words are predicted using another RNN known as decoder. Recently, various models have been proposed which replace the RNN encoder with a convolutional neural network (CNN. In this paper, we propose to augment the standard RNN encoder in NMT with additional convolutional layers in order to capture wider context in the encoder output. Experiments on English to German translation demonstrate that our approach can achieve significant improvements over a standard RNN-based baseline.
Algorithms for Zero-Dimensional Ideals Using Linear Recurrent Sequences
DEFF Research Database (Denmark)
Neiger, Vincent; Rahkooy, Hamid; Schost, Éric
2017-01-01
Inspired by Faugére and Mou´s sparse FGLM algorithm, we show how using linear recurrent multi-dimensional sequences can allow one to perform operations such as the primary decomposition of an ideal, by computing of the annihilator of one or several such sequences.......Inspired by Faugére and Mou´s sparse FGLM algorithm, we show how using linear recurrent multi-dimensional sequences can allow one to perform operations such as the primary decomposition of an ideal, by computing of the annihilator of one or several such sequences....
Synthesis of recurrent neural networks for dynamical system simulation.
Trischler, Adam P; D'Eleuterio, Gabriele M T
2016-08-01
We review several of the most widely used techniques for training recurrent neural networks to approximate dynamical systems, then describe a novel algorithm for this task. The algorithm is based on an earlier theoretical result that guarantees the quality of the network approximation. We show that a feedforward neural network can be trained on the vector-field representation of a given dynamical system using backpropagation, then recast it as a recurrent network that replicates the original system's dynamics. After detailing this algorithm and its relation to earlier approaches, we present numerical examples that demonstrate its capabilities. One of the distinguishing features of our approach is that both the original dynamical systems and the recurrent networks that simulate them operate in continuous time. Copyright © 2016 Elsevier Ltd. All rights reserved.
Recursive Bayesian recurrent neural networks for time-series modeling.
Mirikitani, Derrick T; Nikolaev, Nikolay
2010-02-01
This paper develops a probabilistic approach to recursive second-order training of recurrent neural networks (RNNs) for improved time-series modeling. A general recursive Bayesian Levenberg-Marquardt algorithm is derived to sequentially update the weights and the covariance (Hessian) matrix. The main strengths of the approach are a principled handling of the regularization hyperparameters that leads to better generalization, and stable numerical performance. The framework involves the adaptation of a noise hyperparameter and local weight prior hyperparameters, which represent the noise in the data and the uncertainties in the model parameters. Experimental investigations using artificial and real-world data sets show that RNNs equipped with the proposed approach outperform standard real-time recurrent learning and extended Kalman training algorithms for recurrent networks, as well as other contemporary nonlinear neural models, on time-series modeling.
SORN: a self-organizing recurrent neural network
Directory of Open Access Journals (Sweden)
Andreea Lazar
2009-10-01
Full Text Available Understanding the dynamics of recurrent neural networks is crucial for explaining how the brain processes information. In the neocortex, a range of different plasticity mechanisms are shaping recurrent networks into effective information processing circuits that learn appropriate representations for time-varying sensory stimuli. However, it has been difficult to mimic these abilities in artificial neural network models. Here we introduce SORN, a self-organizing recurrent network. It combines three distinct forms of local plasticity to learn spatio-temporal patterns in its input while maintaining its dynamics in a healthy regime suitable for learning. The SORN learns to encode information in the form of trajectories through its high-dimensional state space reminiscent of recent biological findings on cortical coding. All three forms of plasticity are shown to be essential for the network's success.
Relation Classification via Recurrent Neural Network
Zhang, Dongxu; Wang, Dong
2015-01-01
Deep learning has gained much success in sentence-level relation classification. For example, convolutional neural networks (CNN) have delivered competitive performance without much effort on feature engineering as the conventional pattern-based methods. Thus a lot of works have been produced based on CNN structures. However, a key issue that has not been well addressed by the CNN-based method is the lack of capability to learn temporal features, especially long-distance dependency between no...
Using a multi-state recurrent neural network to optimize loading patterns in BWRs
International Nuclear Information System (INIS)
Ortiz, Juan Jose; Requena, Ignacio
2004-01-01
A Multi-State Recurrent Neural Network is used to optimize Loading Patterns (LP) in BWRs. We have proposed an energy function that depends on fuel assembly positions and their nuclear cross sections to carry out optimisation. Multi-State Recurrent Neural Networks creates LPs that satisfy the Radial Power Peaking Factor and maximize the effective multiplication factor at the Beginning of the Cycle, and also satisfy the Minimum Critical Power Ratio and Maximum Linear Heat Generation Rate at the End of the Cycle, thereby maximizing the effective multiplication factor. In order to evaluate the LPs, we have used a trained back-propagation neural network to predict the parameter values, instead of using a reactor core simulator, which saved considerable computation time in the search process. We applied this method to find optimal LPs for five cycles of Laguna Verde Nuclear Power Plant (LVNPP) in Mexico
Fractional Hopfield Neural Networks: Fractional Dynamic Associative Recurrent Neural Networks.
Pu, Yi-Fei; Yi, Zhang; Zhou, Ji-Liu
2017-10-01
This paper mainly discusses a novel conceptual framework: fractional Hopfield neural networks (FHNN). As is commonly known, fractional calculus has been incorporated into artificial neural networks, mainly because of its long-term memory and nonlocality. Some researchers have made interesting attempts at fractional neural networks and gained competitive advantages over integer-order neural networks. Therefore, it is naturally makes one ponder how to generalize the first-order Hopfield neural networks to the fractional-order ones, and how to implement FHNN by means of fractional calculus. We propose to introduce a novel mathematical method: fractional calculus to implement FHNN. First, we implement fractor in the form of an analog circuit. Second, we implement FHNN by utilizing fractor and the fractional steepest descent approach, construct its Lyapunov function, and further analyze its attractors. Third, we perform experiments to analyze the stability and convergence of FHNN, and further discuss its applications to the defense against chip cloning attacks for anticounterfeiting. The main contribution of our work is to propose FHNN in the form of an analog circuit by utilizing a fractor and the fractional steepest descent approach, construct its Lyapunov function, prove its Lyapunov stability, analyze its attractors, and apply FHNN to the defense against chip cloning attacks for anticounterfeiting. A significant advantage of FHNN is that its attractors essentially relate to the neuron's fractional order. FHNN possesses the fractional-order-stability and fractional-order-sensitivity characteristics.
A two-layer recurrent neural network for nonsmooth convex optimization problems.
Qin, Sitian; Xue, Xiaoping
2015-06-01
In this paper, a two-layer recurrent neural network is proposed to solve the nonsmooth convex optimization problem subject to convex inequality and linear equality constraints. Compared with existing neural network models, the proposed neural network has a low model complexity and avoids penalty parameters. It is proved that from any initial point, the state of the proposed neural network reaches the equality feasible region in finite time and stays there thereafter. Moreover, the state is unique if the initial point lies in the equality feasible region. The equilibrium point set of the proposed neural network is proved to be equivalent to the Karush-Kuhn-Tucker optimality set of the original optimization problem. It is further proved that the equilibrium point of the proposed neural network is stable in the sense of Lyapunov. Moreover, from any initial point, the state is proved to be convergent to an equilibrium point of the proposed neural network. Finally, as applications, the proposed neural network is used to solve nonlinear convex programming with linear constraints and L1 -norm minimization problems.
Iterative free-energy optimization for recurrent neural networks (INFERNO)
2017-01-01
The intra-parietal lobe coupled with the Basal Ganglia forms a working memory that demonstrates strong planning capabilities for generating robust yet flexible neuronal sequences. Neurocomputational models however, often fails to control long range neural synchrony in recurrent spiking networks due to spontaneous activity. As a novel framework based on the free-energy principle, we propose to see the problem of spikes’ synchrony as an optimization problem of the neurons sub-threshold activity for the generation of long neuronal chains. Using a stochastic gradient descent, a reinforcement signal (presumably dopaminergic) evaluates the quality of one input vector to move the recurrent neural network to a desired activity; depending on the error made, this input vector is strengthened to hill-climb the gradient or elicited to search for another solution. This vector can be learned then by one associative memory as a model of the basal-ganglia to control the recurrent neural network. Experiments on habit learning and on sequence retrieving demonstrate the capabilities of the dual system to generate very long and precise spatio-temporal sequences, above two hundred iterations. Its features are applied then to the sequential planning of arm movements. In line with neurobiological theories, we discuss its relevance for modeling the cortico-basal working memory to initiate flexible goal-directed neuronal chains of causation and its relation to novel architectures such as Deep Networks, Neural Turing Machines and the Free-Energy Principle. PMID:28282439
Global robust stability of delayed recurrent neural networks
International Nuclear Information System (INIS)
Cao Jinde; Huang Deshuang; Qu Yuzhong
2005-01-01
This paper is concerned with the global robust stability of a class of delayed interval recurrent neural networks which contain time-invariant uncertain parameters whose values are unknown but bounded in given compact sets. A new sufficient condition is presented for the existence, uniqueness, and global robust stability of equilibria for interval neural networks with time delays by constructing Lyapunov functional and using matrix-norm inequality. An error is corrected in an earlier publication, and an example is given to show the effectiveness of the obtained results
Directory of Open Access Journals (Sweden)
Mattia Rigotti
2010-10-01
Full Text Available Neural activity of behaving animals, especially in the prefrontal cortex, is highly heterogeneous, with selective responses to diverse aspects of the executed task. We propose a general model of recurrent neural networks that perform complex rule-based tasks, and we show that the diversity of neuronal responses plays a fundamental role when the behavioral responses are context dependent. Specifically, we found that when the inner mental states encoding the task rules are represented by stable patterns of neural activity (attractors of the neural dynamics, the neurons must be selective for combinations of sensory stimuli and inner mental states. Such mixed selectivity is easily obtained by neurons that connect with random synaptic strengths both to the recurrent network and to neurons encoding sensory inputs. The number of randomly connected neurons needed to solve a task is on average only three times as large as the number of neurons needed in a network designed ad hoc. Moreover, the number of needed neurons grows only linearly with the number of task-relevant events and mental states, provided that each neuron responds to a large proportion of events (dense/distributed coding. A biologically realistic implementation of the model captures several aspects of the activity recorded from monkeys performing context dependent tasks. Our findings explain the importance of the diversity of neural responses and provide us with simple and general principles for designing attractor neural networks that perform complex computation.
Predicting local field potentials with recurrent neural networks.
Kim, Louis; Harer, Jacob; Rangamani, Akshay; Moran, James; Parks, Philip D; Widge, Alik; Eskandar, Emad; Dougherty, Darin; Chin, Sang Peter
2016-08-01
We present a Recurrent Neural Network using LSTM (Long Short Term Memory) that is capable of modeling and predicting Local Field Potentials. We train and test the network on real data recorded from epilepsy patients. We construct networks that predict multi-channel LFPs for 1, 10, and 100 milliseconds forward in time. Our results show that prediction using LSTM outperforms regression when predicting 10 and 100 millisecond forward in time.
Web server's reliability improvements using recurrent neural networks
DEFF Research Database (Denmark)
Madsen, Henrik; Albu, Rǎzvan-Daniel; Felea, Ioan
2012-01-01
In this paper we describe an interesting approach to error prediction illustrated by experimental results. The application consists of monitoring the activity for the web servers in order to collect the specific data. Predicting an error with severe consequences for the performance of a server (t...... usage, network usage and memory usage. We collect different data sets from monitoring the web server's activity and for each one we predict the server's reliability with the proposed recurrent neural network. © 2012 Taylor & Francis Group...
Parameter estimation in space systems using recurrent neural networks
Parlos, Alexander G.; Atiya, Amir F.; Sunkel, John W.
1991-01-01
The identification of time-varying parameters encountered in space systems is addressed, using artificial neural systems. A hybrid feedforward/feedback neural network, namely a recurrent multilayer perception, is used as the model structure in the nonlinear system identification. The feedforward portion of the network architecture provides its well-known interpolation property, while through recurrency and cross-talk, the local information feedback enables representation of temporal variations in the system nonlinearities. The standard back-propagation-learning algorithm is modified and it is used for both the off-line and on-line supervised training of the proposed hybrid network. The performance of recurrent multilayer perceptron networks in identifying parameters of nonlinear dynamic systems is investigated by estimating the mass properties of a representative large spacecraft. The changes in the spacecraft inertia are predicted using a trained neural network, during two configurations corresponding to the early and late stages of the spacecraft on-orbit assembly sequence. The proposed on-line mass properties estimation capability offers encouraging results, though, further research is warranted for training and testing the predictive capabilities of these networks beyond nominal spacecraft operations.
Prediction of Bladder Cancer Recurrences Using Artificial Neural Networks
Zulueta Guerrero, Ekaitz; Garay, Naiara Telleria; Lopez-Guede, Jose Manuel; Vilches, Borja Ayerdi; Iragorri, Eider Egilegor; Castaños, David Lecumberri; de La Hoz Rastrollo, Ana Belén; Peña, Carlos Pertusa
Even if considerable advances have been made in the field of early diagnosis, there is no simple, cheap and non-invasive method that can be applied to the clinical monitorisation of bladder cancer patients. Moreover, bladder cancer recurrences or the reappearance of the tumour after its surgical resection cannot be predicted in the current clinical setting. In this study, Artificial Neural Networks (ANN) were used to assess how different combinations of classical clinical parameters (stage-grade and age) and two urinary markers (growth factor and pro-inflammatory mediator) could predict post surgical recurrences in bladder cancer patients. Different ANN methods, input parameter combinations and recurrence related output variables were used and the resulting positive and negative prediction rates compared. MultiLayer Perceptron (MLP) was selected as the most predictive model and urinary markers showed the highest sensitivity, predicting correctly 50% of the patients that would recur in a 2 year follow-up period.
Particle Swarm Based Approach of a Real-Time Discrete Neural Identifier for Linear Induction Motors
Directory of Open Access Journals (Sweden)
Alma Y. Alanis
2013-01-01
Full Text Available This paper focusses on a discrete-time neural identifier applied to a linear induction motor (LIM model, whose model is assumed to be unknown. This neural identifier is robust in presence of external and internal uncertainties. The proposed scheme is based on a discrete-time recurrent high-order neural network (RHONN trained with a novel algorithm based on extended Kalman filter (EKF and particle swarm optimization (PSO, using an online series-parallel con figuration. Real-time results are included in order to illustrate the applicability of the proposed scheme.
Recurrent Neural Network for Computing the Drazin Inverse.
Stanimirović, Predrag S; Zivković, Ivan S; Wei, Yimin
2015-11-01
This paper presents a recurrent neural network (RNN) for computing the Drazin inverse of a real matrix in real time. This recurrent neural network (RNN) is composed of n independent parts (subnetworks), where n is the order of the input matrix. These subnetworks can operate concurrently, so parallel and distributed processing can be achieved. In this way, the computational advantages over the existing sequential algorithms can be attained in real-time applications. The RNN defined in this paper is convenient for an implementation in an electronic circuit. The number of neurons in the neural network is the same as the number of elements in the output matrix, which represents the Drazin inverse. The difference between the proposed RNN and the existing ones for the Drazin inverse computation lies in their network architecture and dynamics. The conditions that ensure the stability of the defined RNN as well as its convergence toward the Drazin inverse are considered. In addition, illustrative examples and examples of application to the practical engineering problems are discussed to show the efficacy of the proposed neural network.
A Recurrent Neural Network for Nonlinear Fractional Programming
Directory of Open Access Journals (Sweden)
Quan-Ju Zhang
2012-01-01
Full Text Available This paper presents a novel recurrent time continuous neural network model which performs nonlinear fractional optimization subject to interval constraints on each of the optimization variables. The network is proved to be complete in the sense that the set of optima of the objective function to be minimized with interval constraints coincides with the set of equilibria of the neural network. It is also shown that the network is primal and globally convergent in the sense that its trajectory cannot escape from the feasible region and will converge to an exact optimal solution for any initial point being chosen in the feasible interval region. Simulation results are given to demonstrate further the global convergence and good performance of the proposing neural network for nonlinear fractional programming problems with interval constraints.
Ideomotor feedback control in a recurrent neural network.
Galtier, Mathieu
2015-06-01
The architecture of a neural network controlling an unknown environment is presented. It is based on a randomly connected recurrent neural network from which both perception and action are simultaneously read and fed back. There are two concurrent learning rules implementing a sort of ideomotor control: (i) perception is learned along the principle that the network should predict reliably its incoming stimuli; (ii) action is learned along the principle that the prediction of the network should match a target time series. The coherent behavior of the neural network in its environment is a consequence of the interaction between the two principles. Numerical simulations show a promising performance of the approach, which can be turned into a local and better "biologically plausible" algorithm.
A novel word spotting method based on recurrent neural networks.
Frinken, Volkmar; Fischer, Andreas; Manmatha, R; Bunke, Horst
2012-02-01
Keyword spotting refers to the process of retrieving all instances of a given keyword from a document. In the present paper, a novel keyword spotting method for handwritten documents is described. It is derived from a neural network-based system for unconstrained handwriting recognition. As such it performs template-free spotting, i.e., it is not necessary for a keyword to appear in the training set. The keyword spotting is done using a modification of the CTC Token Passing algorithm in conjunction with a recurrent neural network. We demonstrate that the proposed systems outperform not only a classical dynamic time warping-based approach but also a modern keyword spotting system, based on hidden Markov models. Furthermore, we analyze the performance of the underlying neural networks when using them in a recognition task followed by keyword spotting on the produced transcription. We point out the advantages of keyword spotting when compared to classic text line recognition.
Neural network-based nonlinear model predictive control vs. linear quadratic gaussian control
Cho, C.; Vance, R.; Mardi, N.; Qian, Z.; Prisbrey, K.
1997-01-01
One problem with the application of neural networks to the multivariable control of mineral and extractive processes is determining whether and how to use them. The objective of this investigation was to compare neural network control to more conventional strategies and to determine if there are any advantages in using neural network control in terms of set-point tracking, rise time, settling time, disturbance rejection and other criteria. The procedure involved developing neural network controllers using both historical plant data and simulation models. Various control patterns were tried, including both inverse and direct neural network plant models. These were compared to state space controllers that are, by nature, linear. For grinding and leaching circuits, a nonlinear neural network-based model predictive control strategy was superior to a state space-based linear quadratic gaussian controller. The investigation pointed out the importance of incorporating state space into neural networks by making them recurrent, i.e., feeding certain output state variables into input nodes in the neural network. It was concluded that neural network controllers can have better disturbance rejection, set-point tracking, rise time, settling time and lower set-point overshoot, and it was also concluded that neural network controllers can be more reliable and easy to implement in complex, multivariable plants.
Convolutional neural networks for prostate cancer recurrence prediction
Kumar, Neeraj; Verma, Ruchika; Arora, Ashish; Kumar, Abhay; Gupta, Sanchit; Sethi, Amit; Gann, Peter H.
2017-03-01
Accurate prediction of the treatment outcome is important for cancer treatment planning. We present an approach to predict prostate cancer (PCa) recurrence after radical prostatectomy using tissue images. We used a cohort whose case vs. control (recurrent vs. non-recurrent) status had been determined using post-treatment follow up. Further, to aid the development of novel biomarkers of PCa recurrence, cases and controls were paired based on matching of other predictive clinical variables such as Gleason grade, stage, age, and race. For this cohort, tissue resection microarray with up to four cores per patient was available. The proposed approach is based on deep learning, and its novelty lies in the use of two separate convolutional neural networks (CNNs) - one to detect individual nuclei even in the crowded areas, and the other to classify them. To detect nuclear centers in an image, the first CNN predicts distance transform of the underlying (but unknown) multi-nuclear map from the input HE image. The second CNN classifies the patches centered at nuclear centers into those belonging to cases or controls. Voting across patches extracted from image(s) of a patient yields the probability of recurrence for the patient. The proposed approach gave 0.81 AUC for a sample of 30 recurrent cases and 30 non-recurrent controls, after being trained on an independent set of 80 case-controls pairs. If validated further, such an approach might help in choosing between a combination of treatment options such as active surveillance, radical prostatectomy, radiation, and hormone therapy. It can also generalize to the prediction of treatment outcomes in other cancers.
Model for a flexible motor memory based on a self-active recurrent neural network.
Boström, Kim Joris; Wagner, Heiko; Prieske, Markus; de Lussanet, Marc
2013-10-01
Using recent recurrent network architecture based on the reservoir computing approach, we propose and numerically simulate a model that is focused on the aspects of a flexible motor memory for the storage of elementary movement patterns into the synaptic weights of a neural network, so that the patterns can be retrieved at any time by simple static commands. The resulting motor memory is flexible in that it is capable to continuously modulate the stored patterns. The modulation consists in an approximately linear inter- and extrapolation, generating a large space of possible movements that have not been learned before. A recurrent network of thousand neurons is trained in a manner that corresponds to a realistic exercising scenario, with experimentally measured muscular activations and with kinetic data representing proprioceptive feedback. The network is "self-active" in that it maintains recurrent flow of activation even in the absence of input, a feature that resembles the "resting-state activity" found in the human and animal brain. The model involves the concept of "neural outsourcing" which amounts to the permanent shifting of computational load from higher to lower-level neural structures, which might help to explain why humans are able to execute learned skills in a fluent and flexible manner without the need for attention to the details of the movement. Copyright © 2013 Elsevier B.V. All rights reserved.
Delay-slope-dependent stability results of recurrent neural networks.
Li, Tao; Zheng, Wei Xing; Lin, Chong
2011-12-01
By using the fact that the neuron activation functions are sector bounded and nondecreasing, this brief presents a new method, named the delay-slope-dependent method, for stability analysis of a class of recurrent neural networks with time-varying delays. This method includes more information on the slope of neuron activation functions and fewer matrix variables in the constructed Lyapunov-Krasovskii functional. Then some improved delay-dependent stability criteria with less computational burden and conservatism are obtained. Numerical examples are given to illustrate the effectiveness and the benefits of the proposed method.
Very deep recurrent convolutional neural network for object recognition
Brahimi, Sourour; Ben Aoun, Najib; Ben Amar, Chokri
2017-03-01
In recent years, Computer vision has become a very active field. This field includes methods for processing, analyzing, and understanding images. The most challenging problems in computer vision are image classification and object recognition. This paper presents a new approach for object recognition task. This approach exploits the success of the Very Deep Convolutional Neural Network for object recognition. In fact, it improves the convolutional layers by adding recurrent connections. This proposed approach was evaluated on two object recognition benchmarks: Pascal VOC 2007 and CIFAR-10. The experimental results prove the efficiency of our method in comparison with the state of the art methods.
Liu, Qingshan; Guo, Zhishan; Wang, Jun
2012-02-01
In this paper, a one-layer recurrent neural network is proposed for solving pseudoconvex optimization problems subject to linear equality and bound constraints. Compared with the existing neural networks for optimization (e.g., the projection neural networks), the proposed neural network is capable of solving more general pseudoconvex optimization problems with equality and bound constraints. Moreover, it is capable of solving constrained fractional programming problems as a special case. The convergence of the state variables of the proposed neural network to achieve solution optimality is guaranteed as long as the designed parameters in the model are larger than the derived lower bounds. Numerical examples with simulation results illustrate the effectiveness and characteristics of the proposed neural network. In addition, an application for dynamic portfolio optimization is discussed. Copyright © 2011 Elsevier Ltd. All rights reserved.
Liu, Qingshan; Dang, Chuangyin; Cao, Jinde
2010-07-01
In this paper, based on a one-neuron recurrent neural network, a novel k-winners-take-all ( k -WTA) network is proposed. Finite time convergence of the proposed neural network is proved using the Lyapunov method. The k-WTA operation is first converted equivalently into a linear programming problem. Then, a one-neuron recurrent neural network is proposed to get the kth or (k+1)th largest inputs of the k-WTA problem. Furthermore, a k-WTA network is designed based on the proposed neural network to perform the k-WTA operation. Compared with the existing k-WTA networks, the proposed network has simple structure and finite time convergence. In addition, simulation results on numerical examples show the effectiveness and performance of the proposed k-WTA network.
Learning text representation using recurrent convolutional neural network with highway layers
Wen, Ying; Zhang, Weinan; Luo, Rui; Wang, Jun
2016-01-01
Recently, the rapid development of word embedding and neural networks has brought new inspiration to various NLP and IR tasks. In this paper, we describe a staged hybrid model combining Recurrent Convolutional Neural Networks (RCNN) with highway layers. The highway network module is incorporated in the middle takes the output of the bi-directional Recurrent Neural Network (Bi-RNN) module in the first stage and provides the Convolutional Neural Network (CNN) module in the last stage with the i...
New results on global exponential stability of recurrent neural networks with time-varying delays
International Nuclear Information System (INIS)
Xu Shengyuan; Chu Yuming; Lu Junwei
2006-01-01
This Letter provides new sufficient conditions for the existence, uniqueness and global exponential stability of the equilibrium point of recurrent neural networks with time-varying delays by employing Lyapunov functions and using the Halanay inequality. The time-varying delays are not necessarily differentiable. Both Lipschitz continuous activation functions and monotone nondecreasing activation functions are considered. The derived stability criteria are expressed in terms of linear matrix inequalities (LMIs), which can be checked easily by resorting to recently developed algorithms solving LMIs. Furthermore, the proposed stability results are less conservative than some previous ones in the literature, which is demonstrated via some numerical examples
New results on global exponential stability of recurrent neural networks with time-varying delays
Energy Technology Data Exchange (ETDEWEB)
Xu Shengyuan [Department of Automation, Nanjing University of Science and Technology, Nanjing 210094 (China)]. E-mail: syxu02@yahoo.com.cn; Chu Yuming [Department of Mathematics, Huzhou Teacher' s College, Huzhou, Zhejiang 313000 (China); Lu Junwei [School of Electrical and Automation Engineering, Nanjing Normal University, 78 Bancang Street, Nanjing, 210042 (China)
2006-04-03
This Letter provides new sufficient conditions for the existence, uniqueness and global exponential stability of the equilibrium point of recurrent neural networks with time-varying delays by employing Lyapunov functions and using the Halanay inequality. The time-varying delays are not necessarily differentiable. Both Lipschitz continuous activation functions and monotone nondecreasing activation functions are considered. The derived stability criteria are expressed in terms of linear matrix inequalities (LMIs), which can be checked easily by resorting to recently developed algorithms solving LMIs. Furthermore, the proposed stability results are less conservative than some previous ones in the literature, which is demonstrated via some numerical examples.
Classification of conductance traces with recurrent neural networks
Lauritzen, Kasper P.; Magyarkuti, András; Balogh, Zoltán; Halbritter, András; Solomon, Gemma C.
2018-02-01
We present a new automated method for structural classification of the traces obtained in break junction experiments. Using recurrent neural networks trained on the traces of minimal cross-sectional area in molecular dynamics simulations, we successfully separate the traces into two classes: point contact or nanowire. This is done without any assumptions about the expected features of each class. The trained neural network is applied to experimental break junction conductance traces, and it separates the classes as well as the previously used experimental methods. The effect of using partial conductance traces is explored, and we show that the method performs equally well using full or partial traces (as long as the trace just prior to breaking is included). When only the initial part of the trace is included, the results are still better than random chance. Finally, we show that the neural network classification method can be used to classify experimental conductance traces without using simulated results for training, but instead training the network on a few representative experimental traces. This offers a tool to recognize some characteristic motifs of the traces, which can be hard to find by simple data selection algorithms.
Tuning Recurrent Neural Networks for Recognizing Handwritten Arabic Words
Qaralleh, Esam
2013-10-01
Artificial neural networks have the abilities to learn by example and are capable of solving problems that are hard to solve using ordinary rule-based programming. They have many design parameters that affect their performance such as the number and sizes of the hidden layers. Large sizes are slow and small sizes are generally not accurate. Tuning the neural network size is a hard task because the design space is often large and training is often a long process. We use design of experiments techniques to tune the recurrent neural network used in an Arabic handwriting recognition system. We show that best results are achieved with three hidden layers and two subsampling layers. To tune the sizes of these five layers, we use fractional factorial experiment design to limit the number of experiments to a feasible number. Moreover, we replicate the experiment configuration multiple times to overcome the randomness in the training process. The accuracy and time measurements are analyzed and modeled. The two models are then used to locate network sizes that are on the Pareto optimal frontier. The approach described in this paper reduces the label error from 26.2% to 19.8%.
A modular architecture for transparent computation in recurrent neural networks.
Carmantini, Giovanni S; Beim Graben, Peter; Desroches, Mathieu; Rodrigues, Serafim
2017-01-01
Computation is classically studied in terms of automata, formal languages and algorithms; yet, the relation between neural dynamics and symbolic representations and operations is still unclear in traditional eliminative connectionism. Therefore, we suggest a unique perspective on this central issue, to which we would like to refer as transparent connectionism, by proposing accounts of how symbolic computation can be implemented in neural substrates. In this study we first introduce a new model of dynamics on a symbolic space, the versatile shift, showing that it supports the real-time simulation of a range of automata. We then show that the Gödelization of versatile shifts defines nonlinear dynamical automata, dynamical systems evolving on a vectorial space. Finally, we present a mapping between nonlinear dynamical automata and recurrent artificial neural networks. The mapping defines an architecture characterized by its granular modularity, where data, symbolic operations and their control are not only distinguishable in activation space, but also spatially localizable in the network itself, while maintaining a distributed encoding of symbolic representations. The resulting networks simulate automata in real-time and are programmed directly, in the absence of network training. To discuss the unique characteristics of the architecture and their consequences, we present two examples: (i) the design of a Central Pattern Generator from a finite-state locomotive controller, and (ii) the creation of a network simulating a system of interactive automata that supports the parsing of garden-path sentences as investigated in psycholinguistics experiments. Copyright © 2016 Elsevier Ltd. All rights reserved.
Region stability analysis and tracking control of memristive recurrent neural network.
Bao, Gang; Zeng, Zhigang; Shen, Yanjun
2018-02-01
Memristor is firstly postulated by Leon Chua and realized by Hewlett-Packard (HP) laboratory. Research results show that memristor can be used to simulate the synapses of neurons. This paper presents a class of recurrent neural network with HP memristors. Firstly, it shows that memristive recurrent neural network has more compound dynamics than the traditional recurrent neural network by simulations. Then it derives that n dimensional memristive recurrent neural network is composed of [Formula: see text] sub neural networks which do not have a common equilibrium point. By designing the tracking controller, it can make memristive neural network being convergent to the desired sub neural network. At last, two numerical examples are given to verify the validity of our result. Copyright © 2017 Elsevier Ltd. All rights reserved.
Recurrent Neural Network Approach Based on the Integral Representation of the Drazin Inverse.
Stanimirović, Predrag S; Živković, Ivan S; Wei, Yimin
2015-10-01
In this letter, we present the dynamical equation and corresponding artificial recurrent neural network for computing the Drazin inverse for arbitrary square real matrix, without any restriction on its eigenvalues. Conditions that ensure the stability of the defined recurrent neural network as well as its convergence toward the Drazin inverse are considered. Several illustrative examples present the results of computer simulations.
Cascaded bidirectional recurrent neural networks for protein secondary structure prediction.
Chen, Jinmiao; Chaudhari, Narendra
2007-01-01
Protein secondary structure (PSS) prediction is an important topic in bioinformatics. Our study on a large set of non-homologous proteins shows that long-range interactions commonly exist and negatively affect PSS prediction. Besides, we also reveal strong correlations between secondary structure (SS) elements. In order to take into account the long-range interactions and SS-SS correlations, we propose a novel prediction system based on cascaded bidirectional recurrent neural network (BRNN). We compare the cascaded BRNN against another two BRNN architectures, namely the original BRNN architecture used for speech recognition as well as Pollastri's BRNN that was proposed for PSS prediction. Our cascaded BRNN achieves an overall three state accuracy Q3 of 74.38\\%, and reaches a high Segment OVerlap (SOV) of 66.0455. It outperforms the original BRNN and Pollastri's BRNN in both Q3 and SOV. Specifically, it improves the SOV score by 4-6%.
Directory of Open Access Journals (Sweden)
Mehrshad Salmasi
2012-07-01
Full Text Available Active noise control is based on the destructive interference between the primary noise and generated noise from the secondary source. An antinoise of equal amplitude and opposite phase is generated and combined with the primary noise. In this paper, performance of the neural networks is evaluated in active cancellation of sound noise. For this reason, feedforward and recurrent neural networks are designed and trained. After training, performance of the feedforwrad and recurrent networks in noise attenuation are compared. We use Elman network as a recurrent neural network. For simulations, noise signals from a SPIB database are used. In order to compare the networks appropriately, equal number of layers and neurons are considered for the networks. Moreover, training and test samples are similar. Simulation results show that feedforward and recurrent neural networks present good performance in noise cancellation. As it is seen, the ability of recurrent neural network in noise attenuation is better than feedforward network.
Natural Language Video Description using Deep Recurrent Neural Networks
2015-11-23
ht = f (Wxhxt + Whhht−1) (2.1) zt = g(Wzhht) (2.2) where f and g are element-wise non-linear functions such as a sigmoid or hyperbolic tan - gent, xt...space. arXiv preprint arXiv:1301.3781, 2013. 22 [68] Volodymyr Mnih, Nicolas Heess, Alex Graves, et al. Recurrent models of visual attention. In...2010. 2 36 Bibliography [107] Li Yao, Atousa Torabi, Kyunghyun Cho, Nicolas Ballas, Christopher Pal, Hugo Larochelle, and Aaron Courville. Describing
Implementation of neural network based non-linear predictive
DEFF Research Database (Denmark)
Sørensen, Paul Haase; Nørgård, Peter Magnus; Ravn, Ole
1998-01-01
The paper describes a control method for non-linear systems based on generalized predictive control. Generalized predictive control (GPC) was developed to control linear systems including open loop unstable and non-minimum phase systems, but has also been proposed extended for the control of non......-linear systems. GPC is model-based and in this paper we propose the use of a neural network for the modeling of the system. Based on the neural network model a controller with extended control horizon is developed and the implementation issues are discussed, with particular emphasis on an efficient Quasi......-Newton optimization algorithm. The performance is demonstrated on a pneumatic servo system....
Recurrent Neural Networks for Multivariate Time Series with Missing Values.
Che, Zhengping; Purushotham, Sanjay; Cho, Kyunghyun; Sontag, David; Liu, Yan
2018-04-17
Multivariate time series data in practical applications, such as health care, geoscience, and biology, are characterized by a variety of missing values. In time series prediction and other related tasks, it has been noted that missing values and their missing patterns are often correlated with the target labels, a.k.a., informative missingness. There is very limited work on exploiting the missing patterns for effective imputation and improving prediction performance. In this paper, we develop novel deep learning models, namely GRU-D, as one of the early attempts. GRU-D is based on Gated Recurrent Unit (GRU), a state-of-the-art recurrent neural network. It takes two representations of missing patterns, i.e., masking and time interval, and effectively incorporates them into a deep model architecture so that it not only captures the long-term temporal dependencies in time series, but also utilizes the missing patterns to achieve better prediction results. Experiments of time series classification tasks on real-world clinical datasets (MIMIC-III, PhysioNet) and synthetic datasets demonstrate that our models achieve state-of-the-art performance and provide useful insights for better understanding and utilization of missing values in time series analysis.
Deep Recurrent Neural Networks for Human Activity Recognition
Directory of Open Access Journals (Sweden)
Abdulmajid Murad
2017-11-01
Full Text Available Adopting deep learning methods for human activity recognition has been effective in extracting discriminative features from raw input sequences acquired from body-worn sensors. Although human movements are encoded in a sequence of successive samples in time, typical machine learning methods perform recognition tasks without exploiting the temporal correlations between input data samples. Convolutional neural networks (CNNs address this issue by using convolutions across a one-dimensional temporal sequence to capture dependencies among input data. However, the size of convolutional kernels restricts the captured range of dependencies between data samples. As a result, typical models are unadaptable to a wide range of activity-recognition configurations and require fixed-length input windows. In this paper, we propose the use of deep recurrent neural networks (DRNNs for building recognition models that are capable of capturing long-range dependencies in variable-length input sequences. We present unidirectional, bidirectional, and cascaded architectures based on long short-term memory (LSTM DRNNs and evaluate their effectiveness on miscellaneous benchmark datasets. Experimental results show that our proposed models outperform methods employing conventional machine learning, such as support vector machine (SVM and k-nearest neighbors (KNN. Additionally, the proposed models yield better performance than other deep learning techniques, such as deep believe networks (DBNs and CNNs.
Drawing and Recognizing Chinese Characters with Recurrent Neural Network.
Zhang, Xu-Yao; Yin, Fei; Zhang, Yan-Ming; Liu, Cheng-Lin; Bengio, Yoshua
2018-04-01
Recent deep learning based approaches have achieved great success on handwriting recognition. Chinese characters are among the most widely adopted writing systems in the world. Previous research has mainly focused on recognizing handwritten Chinese characters. However, recognition is only one aspect for understanding a language, another challenging and interesting task is to teach a machine to automatically write (pictographic) Chinese characters. In this paper, we propose a framework by using the recurrent neural network (RNN) as both a discriminative model for recognizing Chinese characters and a generative model for drawing (generating) Chinese characters. To recognize Chinese characters, previous methods usually adopt the convolutional neural network (CNN) models which require transforming the online handwriting trajectory into image-like representations. Instead, our RNN based approach is an end-to-end system which directly deals with the sequential structure and does not require any domain-specific knowledge. With the RNN system (combining an LSTM and GRU), state-of-the-art performance can be achieved on the ICDAR-2013 competition database. Furthermore, under the RNN framework, a conditional generative model with character embedding is proposed for automatically drawing recognizable Chinese characters. The generated characters (in vector format) are human-readable and also can be recognized by the discriminative RNN model with high accuracy. Experimental results verify the effectiveness of using RNNs as both generative and discriminative models for the tasks of drawing and recognizing Chinese characters.
Recurrent Neural Networks to Correct Satellite Image Classification Maps
Maggiori, Emmanuel; Charpiat, Guillaume; Tarabalka, Yuliya; Alliez, Pierre
2017-09-01
While initially devised for image categorization, convolutional neural networks (CNNs) are being increasingly used for the pixelwise semantic labeling of images. However, the proper nature of the most common CNN architectures makes them good at recognizing but poor at localizing objects precisely. This problem is magnified in the context of aerial and satellite image labeling, where a spatially fine object outlining is of paramount importance. Different iterative enhancement algorithms have been presented in the literature to progressively improve the coarse CNN outputs, seeking to sharpen object boundaries around real image edges. However, one must carefully design, choose and tune such algorithms. Instead, our goal is to directly learn the iterative process itself. For this, we formulate a generic iterative enhancement process inspired from partial differential equations, and observe that it can be expressed as a recurrent neural network (RNN). Consequently, we train such a network from manually labeled data for our enhancement task. In a series of experiments we show that our RNN effectively learns an iterative process that significantly improves the quality of satellite image classification maps.
Deep Recurrent Neural Networks for Human Activity Recognition.
Murad, Abdulmajid; Pyun, Jae-Young
2017-11-06
Adopting deep learning methods for human activity recognition has been effective in extracting discriminative features from raw input sequences acquired from body-worn sensors. Although human movements are encoded in a sequence of successive samples in time, typical machine learning methods perform recognition tasks without exploiting the temporal correlations between input data samples. Convolutional neural networks (CNNs) address this issue by using convolutions across a one-dimensional temporal sequence to capture dependencies among input data. However, the size of convolutional kernels restricts the captured range of dependencies between data samples. As a result, typical models are unadaptable to a wide range of activity-recognition configurations and require fixed-length input windows. In this paper, we propose the use of deep recurrent neural networks (DRNNs) for building recognition models that are capable of capturing long-range dependencies in variable-length input sequences. We present unidirectional, bidirectional, and cascaded architectures based on long short-term memory (LSTM) DRNNs and evaluate their effectiveness on miscellaneous benchmark datasets. Experimental results show that our proposed models outperform methods employing conventional machine learning, such as support vector machine (SVM) and k-nearest neighbors (KNN). Additionally, the proposed models yield better performance than other deep learning techniques, such as deep believe networks (DBNs) and CNNs.
Global dissipativity of continuous-time recurrent neural networks with time delay
International Nuclear Information System (INIS)
Liao Xiaoxin; Wang Jun
2003-01-01
This paper addresses the global dissipativity of a general class of continuous-time recurrent neural networks. First, the concepts of global dissipation and global exponential dissipation are defined and elaborated. Next, the sets of global dissipativity and global exponentially dissipativity are characterized using the parameters of recurrent neural network models. In particular, it is shown that the Hopfield network and cellular neural networks with or without time delays are dissipative systems
Exponentially convergent state estimation for delayed switched recurrent neural networks.
Ahn, Choon Ki
2011-11-01
This paper deals with the delay-dependent exponentially convergent state estimation problem for delayed switched neural networks. A set of delay-dependent criteria is derived under which the resulting estimation error system is exponentially stable. It is shown that the gain matrix of the proposed state estimator is characterised in terms of the solution to a set of linear matrix inequalities (LMIs), which can be checked readily by using some standard numerical packages. An illustrative example is given to demonstrate the effectiveness of the proposed state estimator.
International Nuclear Information System (INIS)
Yan, Ji; Bao-Tong, Cui
2010-01-01
In this paper, we have improved delay-dependent stability criteria for recurrent neural networks with a delay varying over a range and Markovian jumping parameters. The criteria improve over some previous ones in that they have fewer matrix variables yet less conservatism. In addition, a numerical example is provided to illustrate the applicability of the result using the linear matrix inequality toolbox in MATLAB. (general)
Neural Generalized Predictive Control of a non-linear Process
DEFF Research Database (Denmark)
Sørensen, Paul Haase; Nørgård, Peter Magnus; Ravn, Ole
1998-01-01
The use of neural network in non-linear control is made difficult by the fact the stability and robustness is not guaranteed and that the implementation in real time is non-trivial. In this paper we introduce a predictive controller based on a neural network model which has promising stability qu...... detail and discuss the implementation difficulties. The neural generalized predictive controller is tested on a pneumatic servo sys-tem.......The use of neural network in non-linear control is made difficult by the fact the stability and robustness is not guaranteed and that the implementation in real time is non-trivial. In this paper we introduce a predictive controller based on a neural network model which has promising stability...... qualities. The controller is a non-linear version of the well-known generalized predictive controller developed in linear control theory. It involves minimization of a cost function which in the present case has to be done numerically. Therefore, we develop the numerical algorithms necessary in substantial...
Non-linear feedback neural networks VLSI implementations and applications
Ansari, Mohd Samar
2014-01-01
This book aims to present a viable alternative to the Hopfield Neural Network (HNN) model for analog computation. It is well known that the standard HNN suffers from problems of convergence to local minima, and requirement of a large number of neurons and synaptic weights. Therefore, improved solutions are needed. The non-linear synapse neural network (NoSyNN) is one such possibility and is discussed in detail in this book. This book also discusses the applications in computationally intensive tasks like graph coloring, ranking, and linear as well as quadratic programming. The material in the book is useful to students, researchers and academician working in the area of analog computation.
Recurrent Neural Network Based Boolean Factor Analysis and its Application to Word Clustering
Czech Academy of Sciences Publication Activity Database
Frolov, A. A.; Húsek, Dušan; Polyakov, P.Y.
2009-01-01
Roč. 20, č. 7 (2009), s. 1073-1086 ISSN 1045-9227 R&D Projects: GA MŠk(CZ) 1M0567 Institutional research plan: CEZ:AV0Z10300504 Keywords : recurrent neural network * Hopfield-like neural network * associative memory * unsupervised learning * neural network architecture * neural network application * statistics * Boolean factor analysis * concepts search * information retrieval Subject RIV: BB - Applied Statistics, Operational Research Impact factor: 2.889, year: 2009
Linear programming based on neural networks for radiotherapy treatment planning
International Nuclear Information System (INIS)
Xingen Wu; Limin Luo
2000-01-01
In this paper, we propose a neural network model for linear programming that is designed to optimize radiotherapy treatment planning (RTP). This kind of neural network can be easily implemented by using a kind of 'neural' electronic system in order to obtain an optimization solution in real time. We first give an introduction to the RTP problem and construct a non-constraint objective function for the neural network model. We adopt a gradient algorithm to minimize the objective function and design the structure of the neural network for RTP. Compared to traditional linear programming methods, this neural network model can reduce the time needed for convergence, the size of problems (i.e., the number of variables to be searched) and the number of extra slack and surplus variables needed. We obtained a set of optimized beam weights that result in a better dose distribution as compared to that obtained using the simplex algorithm under the same initial condition. The example presented in this paper shows that this model is feasible in three-dimensional RTP. (author)
Bennett, C.; Dunne, J. F.; Trimby, S.; Richardson, D.
2017-02-01
A recurrent non-linear autoregressive with exogenous input (NARX) neural network is proposed, and a suitable fully-recurrent training methodology is adapted and tuned, for reconstructing cylinder pressure in multi-cylinder IC engines using measured crank kinematics. This type of indirect sensing is important for cost effective closed-loop combustion control and for On-Board Diagnostics. The challenge addressed is to accurately predict cylinder pressure traces within the cycle under generalisation conditions: i.e. using data not previously seen by the network during training. This involves direct construction and calibration of a suitable inverse crank dynamic model, which owing to singular behaviour at top-dead-centre (TDC), has proved difficult via physical model construction, calibration, and inversion. The NARX architecture is specialised and adapted to cylinder pressure reconstruction, using a fully-recurrent training methodology which is needed because the alternatives are too slow and unreliable for practical network training on production engines. The fully-recurrent Robust Adaptive Gradient Descent (RAGD) algorithm, is tuned initially using synthesised crank kinematics, and then tested on real engine data to assess the reconstruction capability. Real data is obtained from a 1.125 l, 3-cylinder, in-line, direct injection spark ignition (DISI) engine involving synchronised measurements of crank kinematics and cylinder pressure across a range of steady-state speed and load conditions. The paper shows that a RAGD-trained NARX network using both crank velocity and crank acceleration as input information, provides fast and robust training. By using the optimum epoch identified during RAGD training, acceptably accurate cylinder pressures, and especially accurate location-of-peak-pressure, can be reconstructed robustly under generalisation conditions, making it the most practical NARX configuration and recurrent training methodology for use on production engines.
Liu, Qingshan; Cao, Jinde
2010-06-01
Based on the projection operator, a recurrent neural network is proposed for solving extended general variational inequalities (EGVIs). Sufficient conditions are provided to ensure the global convergence of the proposed neural network based on Lyapunov methods. Compared with the existing neural networks for variational inequalities, the proposed neural network is a modified version of the general projection neural network existing in the literature and capable of solving the EGVI problems. In addition, simulation results on numerical examples show the effectiveness and performance of the proposed neural network.
Application of recurrent neural networks for drought projections in California
Le, J. A.; El-Askary, H. M.; Allali, M.; Struppa, D. C.
2017-05-01
We use recurrent neural networks (RNNs) to investigate the complex interactions between the long-term trend in dryness and a projected, short but intense, period of wetness due to the 2015-2016 El Niño. Although it was forecasted that this El Niño season would bring significant rainfall to the region, our long-term projections of the Palmer Z Index (PZI) showed a continuing drought trend, contrasting with the 1998-1999 El Niño event. RNN training considered PZI data during 1896-2006 that was validated against the 2006-2015 period to evaluate the potential of extreme precipitation forecast. We achieved a statistically significant correlation of 0.610 between forecasted and observed PZI on the validation set for a lead time of 1 month. This gives strong confidence to the forecasted precipitation indicator. The 2015-2016 El Niño season proved to be relatively weak as compared with the 1997-1998, with a peak PZI anomaly of 0.242 standard deviations below historical averages, continuing drought conditions.
Recurrent Neural Network Model for Constructive Peptide Design.
Müller, Alex T; Hiss, Jan A; Schneider, Gisbert
2018-02-26
We present a generative long short-term memory (LSTM) recurrent neural network (RNN) for combinatorial de novo peptide design. RNN models capture patterns in sequential data and generate new data instances from the learned context. Amino acid sequences represent a suitable input for these machine-learning models. Generative models trained on peptide sequences could therefore facilitate the design of bespoke peptide libraries. We trained RNNs with LSTM units on pattern recognition of helical antimicrobial peptides and used the resulting model for de novo sequence generation. Of these sequences, 82% were predicted to be active antimicrobial peptides compared to 65% of randomly sampled sequences with the same amino acid distribution as the training set. The generated sequences also lie closer to the training data than manually designed amphipathic helices. The results of this study showcase the ability of LSTM RNNs to construct new amino acid sequences within the applicability domain of the model and motivate their prospective application to peptide and protein design without the need for the exhaustive enumeration of sequence libraries.
Multiplex visibility graphs to investigate recurrent neural network dynamics
Bianchi, Filippo Maria; Livi, Lorenzo; Alippi, Cesare; Jenssen, Robert
2017-03-01
A recurrent neural network (RNN) is a universal approximator of dynamical systems, whose performance often depends on sensitive hyperparameters. Tuning them properly may be difficult and, typically, based on a trial-and-error approach. In this work, we adopt a graph-based framework to interpret and characterize internal dynamics of a class of RNNs called echo state networks (ESNs). We design principled unsupervised methods to derive hyperparameters configurations yielding maximal ESN performance, expressed in terms of prediction error and memory capacity. In particular, we propose to model time series generated by each neuron activations with a horizontal visibility graph, whose topological properties have been shown to be related to the underlying system dynamics. Successively, horizontal visibility graphs associated with all neurons become layers of a larger structure called a multiplex. We show that topological properties of such a multiplex reflect important features of ESN dynamics that can be used to guide the tuning of its hyperparamers. Results obtained on several benchmarks and a real-world dataset of telephone call data records show the effectiveness of the proposed methods.
Spatial Clockwork Recurrent Neural Network for Muscle Perimysium Segmentation.
Xie, Yuanpu; Zhang, Zizhao; Sapkota, Manish; Yang, Lin
2016-10-01
Accurate segmentation of perimysium plays an important role in early diagnosis of many muscle diseases because many diseases contain different perimysium inflammation. However, it remains as a challenging task due to the complex appearance of the perymisum morphology and its ambiguity to the background area. The muscle perimysium also exhibits strong structure spanned in the entire tissue, which makes it difficult for current local patch-based methods to capture this long-range context information. In this paper, we propose a novel spatial clockwork recurrent neural network (spatial CW-RNN) to address those issues. Specifically, we split the entire image into a set of non-overlapping image patches, and the semantic dependencies among them are modeled by the proposed spatial CW-RNN. Our method directly takes the 2D structure of the image into consideration and is capable of encoding the context information of the entire image into the local representation of each patch. Meanwhile, we leverage on the structured regression to assign one prediction mask rather than a single class label to each local patch, which enables both efficient training and testing. We extensively test our method for perimysium segmentation using digitized muscle microscopy images. Experimental results demonstrate the superiority of the novel spatial CW-RNN over other existing state of the arts.
Fast computation with spikes in a recurrent neural network
International Nuclear Information System (INIS)
Jin, Dezhe Z.; Seung, H. Sebastian
2002-01-01
Neural networks with recurrent connections are sometimes regarded as too slow at computation to serve as models of the brain. Here we analytically study a counterexample, a network consisting of N integrate-and-fire neurons with self excitation, all-to-all inhibition, instantaneous synaptic coupling, and constant external driving inputs. When the inhibition and/or excitation are large enough, the network performs a winner-take-all computation for all possible external inputs and initial states of the network. The computation is done very quickly: As soon as the winner spikes once, the computation is completed since no other neurons will spike. For some initial states, the winner is the first neuron to spike, and the computation is done at the first spike of the network. In general, there are M potential winners, corresponding to the top M external inputs. When the external inputs are close in magnitude, M tends to be larger. If M>1, the selection of the actual winner is strongly influenced by the initial states. If a special relation between the excitation and inhibition is satisfied, the network always selects the neuron with the maximum external input as the winner
Low-dimensional recurrent neural network-based Kalman filter for speech enhancement.
Xia, Youshen; Wang, Jun
2015-07-01
This paper proposes a new recurrent neural network-based Kalman filter for speech enhancement, based on a noise-constrained least squares estimate. The parameters of speech signal modeled as autoregressive process are first estimated by using the proposed recurrent neural network and the speech signal is then recovered from Kalman filtering. The proposed recurrent neural network is globally asymptomatically stable to the noise-constrained estimate. Because the noise-constrained estimate has a robust performance against non-Gaussian noise, the proposed recurrent neural network-based speech enhancement algorithm can minimize the estimation error of Kalman filter parameters in non-Gaussian noise. Furthermore, having a low-dimensional model feature, the proposed neural network-based speech enhancement algorithm has a much faster speed than two existing recurrent neural networks-based speech enhancement algorithms. Simulation results show that the proposed recurrent neural network-based speech enhancement algorithm can produce a good performance with fast computation and noise reduction. Copyright © 2015 Elsevier Ltd. All rights reserved.
Cheng, Long; Hou, Zeng-Guang; Lin, Yingzi; Tan, Min; Zhang, Wenjun Chris; Wu, Fang-Xiang
2011-05-01
A recurrent neural network is proposed for solving the non-smooth convex optimization problem with the convex inequality and linear equality constraints. Since the objective function and inequality constraints may not be smooth, the Clarke's generalized gradients of the objective function and inequality constraints are employed to describe the dynamics of the proposed neural network. It is proved that the equilibrium point set of the proposed neural network is equivalent to the optimal solution of the original optimization problem by using the Lagrangian saddle-point theorem. Under weak conditions, the proposed neural network is proved to be stable, and the state of the neural network is convergent to one of its equilibrium points. Compared with the existing neural network models for non-smooth optimization problems, the proposed neural network can deal with a larger class of constraints and is not based on the penalty method. Finally, the proposed neural network is used to solve the identification problem of genetic regulatory networks, which can be transformed into a non-smooth convex optimization problem. The simulation results show the satisfactory identification accuracy, which demonstrates the effectiveness and efficiency of the proposed approach.
Solving differential equations with unknown constitutive relations as recurrent neural networks
Energy Technology Data Exchange (ETDEWEB)
Hagge, Tobias J.; Stinis, Panagiotis; Yeung, Enoch H.; Tartakovsky, Alexandre M.
2017-12-08
We solve a system of ordinary differential equations with an unknown functional form of a sink (reaction rate) term. We assume that the measurements (time series) of state variables are partially available, and use a recurrent neural network to “learn” the reaction rate from this data. This is achieved by including discretized ordinary differential equations as part of a recurrent neural network training problem. We extend TensorFlow’s recurrent neural network architecture to create a simple but scalable and effective solver for the unknown functions, and apply it to a fedbatch bioreactor simulation problem. Use of techniques from recent deep learning literature enables training of functions with behavior manifesting over thousands of time steps. Our networks are structurally similar to recurrent neural networks, but differ in purpose, and require modified training strategies.
Zhao, Haiquan; Zeng, Xiangping; Zhang, Jiashu; Liu, Yangguang; Wang, Xiaomin; Li, Tianrui
2011-01-01
To eliminate nonlinear channel distortion in chaotic communication systems, a novel joint-processing adaptive nonlinear equalizer based on a pipelined recurrent neural network (JPRNN) is proposed, using a modified real-time recurrent learning (RTRL) algorithm. Furthermore, an adaptive amplitude RTRL algorithm is adopted to overcome the deteriorating effect introduced by the nesting process. Computer simulations illustrate that the proposed equalizer outperforms the pipelined recurrent neural network (PRNN) and recurrent neural network (RNN) equalizers. Copyright © 2010 Elsevier Ltd. All rights reserved.
Directory of Open Access Journals (Sweden)
J. Humberto Pérez-Cruz
2012-01-01
Full Text Available This paper deals with the problem of trajectory tracking for a broad class of uncertain nonlinear systems with multiple inputs each one subject to an unknown symmetric deadzone. On the basis of a model of the deadzone as a combination of a linear term and a disturbance-like term, a continuous-time recurrent neural network is directly employed in order to identify the uncertain dynamics. By using a Lyapunov analysis, the exponential convergence of the identification error to a bounded zone is demonstrated. Subsequently, by a proper control law, the state of the neural network is compelled to follow a bounded reference trajectory. This control law is designed in such a way that the singularity problem is conveniently avoided and the exponential convergence to a bounded zone of the difference between the state of the neural identifier and the reference trajectory can be proven. Thus, the exponential convergence of the tracking error to a bounded zone and the boundedness of all closed-loop signals can be guaranteed. One of the main advantages of the proposed strategy is that the controller can work satisfactorily without any specific knowledge of an upper bound for the unmodeled dynamics and/or the disturbance term.
Mehrshad Salmasi; Homayoun Mahdavi-Nasab
2012-01-01
Active noise control is based on the destructive interference between the primary noise and generated noise from the secondary source. An antinoise of equal amplitude and opposite phase is generated and combined with the primary noise. In this paper, performance of the neural networks is evaluated in active cancellation of sound noise. For this reason, feedforward and recurrent neural networks are designed and trained. After training, performance of the feedforwrad and recurrent networks in n...
International Nuclear Information System (INIS)
Liang Jinling; Cao Jinde
2003-01-01
Employing general Halanay inequality, we analyze the global exponential stability of a class of reaction-diffusion recurrent neural networks with time-varying delays. Several new sufficient conditions are obtained to ensure existence, uniqueness and global exponential stability of the equilibrium point of delayed reaction-diffusion recurrent neural networks. The results extend and improve the earlier publications. In addition, an example is given to show the effectiveness of the obtained result
Encoding sensory and motor patterns as time-invariant trajectories in recurrent neural networks.
Goudar, Vishwa; Buonomano, Dean V
2018-03-14
Much of the information the brain processes and stores is temporal in nature-a spoken word or a handwritten signature, for example, is defined by how it unfolds in time. However, it remains unclear how neural circuits encode complex time-varying patterns. We show that by tuning the weights of a recurrent neural network (RNN), it can recognize and then transcribe spoken digits. The model elucidates how neural dynamics in cortical networks may resolve three fundamental challenges: first, encode multiple time-varying sensory and motor patterns as stable neural trajectories; second, generalize across relevant spatial features; third, identify the same stimuli played at different speeds-we show that this temporal invariance emerges because the recurrent dynamics generate neural trajectories with appropriately modulated angular velocities. Together our results generate testable predictions as to how recurrent networks may use different mechanisms to generalize across the relevant spatial and temporal features of complex time-varying stimuli. © 2018, Goudar et al.
Implementation of neural network based non-linear predictive control
DEFF Research Database (Denmark)
Sørensen, Paul Haase; Nørgård, Peter Magnus; Ravn, Ole
1999-01-01
This paper describes a control method for non-linear systems based on generalized predictive control. Generalized predictive control (GPC) was developed to control linear systems, including open-loop unstable and non-minimum phase systems, but has also been proposed to be extended for the control...... of non-linear systems. GPC is model based and in this paper we propose the use of a neural network for the modeling of the system. Based on the neural network model, a controller with extended control horizon is developed and the implementation issues are discussed, with particular emphasis...... on an efficient quasi-Newton algorithm. The performance is demonstrated on a pneumatic servo system....
Liu, Hongjian; Wang, Zidong; Shen, Bo; Alsaadi, Fuad E.
2016-07-01
This paper deals with the robust H∞ state estimation problem for a class of memristive recurrent neural networks with stochastic time-delays. The stochastic time-delays under consideration are governed by a Bernoulli-distributed stochastic sequence. The purpose of the addressed problem is to design the robust state estimator such that the dynamics of the estimation error is exponentially stable in the mean square, and the prescribed ? performance constraint is met. By utilizing the difference inclusion theory and choosing a proper Lyapunov-Krasovskii functional, the existence condition of the desired estimator is derived. Based on it, the explicit expression of the estimator gain is given in terms of the solution to a linear matrix inequality. Finally, a numerical example is employed to demonstrate the effectiveness and applicability of the proposed estimation approach.
Discrete-time recurrent neural networks with time-varying delays: Exponential stability analysis
International Nuclear Information System (INIS)
Liu, Yurong; Wang, Zidong; Serrano, Alan; Liu, Xiaohui
2007-01-01
This Letter is concerned with the analysis problem of exponential stability for a class of discrete-time recurrent neural networks (DRNNs) with time delays. The delay is of the time-varying nature, and the activation functions are assumed to be neither differentiable nor strict monotonic. Furthermore, the description of the activation functions is more general than the recently commonly used Lipschitz conditions. Under such mild conditions, we first prove the existence of the equilibrium point. Then, by employing a Lyapunov-Krasovskii functional, a unified linear matrix inequality (LMI) approach is developed to establish sufficient conditions for the DRNNs to be globally exponentially stable. It is shown that the delayed DRNNs are globally exponentially stable if a certain LMI is solvable, where the feasibility of such an LMI can be easily checked by using the numerically efficient Matlab LMI Toolbox. A simulation example is presented to show the usefulness of the derived LMI-based stability condition
Recurrent Convolutional Neural Networks: A Better Model of Biological Object Recognition.
Spoerer, Courtney J; McClure, Patrick; Kriegeskorte, Nikolaus
2017-01-01
Feedforward neural networks provide the dominant model of how the brain performs visual object recognition. However, these networks lack the lateral and feedback connections, and the resulting recurrent neuronal dynamics, of the ventral visual pathway in the human and non-human primate brain. Here we investigate recurrent convolutional neural networks with bottom-up (B), lateral (L), and top-down (T) connections. Combining these types of connections yields four architectures (B, BT, BL, and BLT), which we systematically test and compare. We hypothesized that recurrent dynamics might improve recognition performance in the challenging scenario of partial occlusion. We introduce two novel occluded object recognition tasks to test the efficacy of the models, digit clutter (where multiple target digits occlude one another) and digit debris (where target digits are occluded by digit fragments). We find that recurrent neural networks outperform feedforward control models (approximately matched in parametric complexity) at recognizing objects, both in the absence of occlusion and in all occlusion conditions. Recurrent networks were also found to be more robust to the inclusion of additive Gaussian noise. Recurrent neural networks are better in two respects: (1) they are more neurobiologically realistic than their feedforward counterparts; (2) they are better in terms of their ability to recognize objects, especially under challenging conditions. This work shows that computer vision can benefit from using recurrent convolutional architectures and suggests that the ubiquitous recurrent connections in biological brains are essential for task performance.
A One-Layer Recurrent Neural Network for Constrained Complex-Variable Convex Optimization.
Qin, Sitian; Feng, Jiqiang; Song, Jiahui; Wen, Xingnan; Xu, Chen
2018-03-01
In this paper, based on calculus and penalty method, a one-layer recurrent neural network is proposed for solving constrained complex-variable convex optimization. It is proved that for any initial point from a given domain, the state of the proposed neural network reaches the feasible region in finite time and converges to an optimal solution of the constrained complex-variable convex optimization finally. In contrast to existing neural networks for complex-variable convex optimization, the proposed neural network has a lower model complexity and better convergence. Some numerical examples and application are presented to substantiate the effectiveness of the proposed neural network.
Directory of Open Access Journals (Sweden)
Suhartono Suhartono
2009-07-01
Full Text Available Neural network (NN is one of many method used to predict the electricity consumption per hour in many countries. NN method which is used in many previous studies is Feed-Forward Neural Network (FFNN or Autoregressive Neural Network(AR-NN. AR-NN model is not able to capture and explain the effect of moving average (MA order on a time series of data. This research was conducted with the purpose of reviewing the application of other types of NN, that is Elman-Recurrent Neural Network (Elman-RNN which could explain MA order effect and compare the result of prediction accuracy with multiple seasonal ARIMA (Autoregressive Integrated Moving Average models. As a case study, we used data electricity consumption per hour in Mengare Gresik. Result of analysis showed that the best of double seasonal Arima models suited to short-term forecasting in the case study data is ARIMA([1,2,3,4,6,7,9,10,14,21,33],1,8(0,1,124 (1,1,0168. This model produces a white noise residuals, but it does not have a normal distribution due to suspected outlier. Outlier detection in iterative produce 14 innovation outliers. There are 4 inputs of Elman-RNN network that were examined and tested for forecasting the data, the input according to lag Arima, input such as lag Arima plus 14 dummy outlier, inputs are the lag-multiples of 24 up to lag 480, and the inputs are lag 1 and lag multiples of 24+1. All of four network uses one hidden layer with tangent sigmoid activation function and one output with a linear function. The result of comparative forecast accuracy through value of MAPE out-sample showed that the fourth networks, namely Elman-RNN (22, 3, 1, is the best model for forecasting electricity consumption per hour in short term in Mengare Gresik.
Multi-step-prediction of chaotic time series based on co-evolutionary recurrent neural network
International Nuclear Information System (INIS)
Ma Qianli; Zheng Qilun; Peng Hong; Qin Jiangwei; Zhong Tanwei
2008-01-01
This paper proposes a co-evolutionary recurrent neural network (CERNN) for the multi-step-prediction of chaotic time series, it estimates the proper parameters of phase space reconstruction and optimizes the structure of recurrent neural networks by co-evolutionary strategy. The searching space was separated into two subspaces and the individuals are trained in a parallel computational procedure. It can dynamically combine the embedding method with the capability of recurrent neural network to incorporate past experience due to internal recurrence. The effectiveness of CERNN is evaluated by using three benchmark chaotic time series data sets: the Lorenz series, Mackey-Glass series and real-world sun spot series. The simulation results show that CERNN improves the performances of multi-step-prediction of chaotic time series
Directory of Open Access Journals (Sweden)
Juan Andres Laura
2018-03-01
Full Text Available In recent studies Recurrent Neural Networks were used for generative processes and their surprising performance can be explained by their ability to create good predictions. In addition, Data Compression is also based on prediction. What the problem comes down to is whether a data compressor could be used to perform as well as recurrent neural networks in the natural language processing tasks of sentiment analysis and automatic text generation. If this is possible, then the problem comes down to determining if a compression algorithm is even more intelligent than a neural network in such tasks. In our journey, a fundamental difference between a Data Compression Algorithm and Recurrent Neural Networks has been discovered.
Transiently chaotic neural networks with piecewise linear output functions
Energy Technology Data Exchange (ETDEWEB)
Chen, S.-S. [Department of Mathematics, National Taiwan Normal University, Taipei, Taiwan (China); Shih, C.-W. [Department of Applied Mathematics, National Chiao Tung University, 1001 Ta-Hsueh Road, Hsinchu, Taiwan (China)], E-mail: cwshih@math.nctu.edu.tw
2009-01-30
Admitting both transient chaotic phase and convergent phase, the transiently chaotic neural network (TCNN) provides superior performance than the classical networks in solving combinatorial optimization problems. We derive concrete parameter conditions for these two essential dynamic phases of the TCNN with piecewise linear output function. The confirmation for chaotic dynamics of the system results from a successful application of the Marotto theorem which was recently clarified. Numerical simulation on applying the TCNN with piecewise linear output function is carried out to find the optimal solution of a travelling salesman problem. It is demonstrated that the performance is even better than the previous TCNN model with logistic output function.
An Attractor-Based Complexity Measurement for Boolean Recurrent Neural Networks
Cabessa, Jérémie; Villa, Alessandro E. P.
2014-01-01
We provide a novel refined attractor-based complexity measurement for Boolean recurrent neural networks that represents an assessment of their computational power in terms of the significance of their attractor dynamics. This complexity measurement is achieved by first proving a computational equivalence between Boolean recurrent neural networks and some specific class of -automata, and then translating the most refined classification of -automata to the Boolean neural network context. As a result, a hierarchical classification of Boolean neural networks based on their attractive dynamics is obtained, thus providing a novel refined attractor-based complexity measurement for Boolean recurrent neural networks. These results provide new theoretical insights to the computational and dynamical capabilities of neural networks according to their attractive potentialities. An application of our findings is illustrated by the analysis of the dynamics of a simplified model of the basal ganglia-thalamocortical network simulated by a Boolean recurrent neural network. This example shows the significance of measuring network complexity, and how our results bear new founding elements for the understanding of the complexity of real brain circuits. PMID:24727866
Entity recognition from clinical texts via recurrent neural network.
Liu, Zengjian; Yang, Ming; Wang, Xiaolong; Chen, Qingcai; Tang, Buzhou; Wang, Zhe; Xu, Hua
2017-07-05
Entity recognition is one of the most primary steps for text analysis and has long attracted considerable attention from researchers. In the clinical domain, various types of entities, such as clinical entities and protected health information (PHI), widely exist in clinical texts. Recognizing these entities has become a hot topic in clinical natural language processing (NLP), and a large number of traditional machine learning methods, such as support vector machine and conditional random field, have been deployed to recognize entities from clinical texts in the past few years. In recent years, recurrent neural network (RNN), one of deep learning methods that has shown great potential on many problems including named entity recognition, also has been gradually used for entity recognition from clinical texts. In this paper, we comprehensively investigate the performance of LSTM (long-short term memory), a representative variant of RNN, on clinical entity recognition and protected health information recognition. The LSTM model consists of three layers: input layer - generates representation of each word of a sentence; LSTM layer - outputs another word representation sequence that captures the context information of each word in this sentence; Inference layer - makes tagging decisions according to the output of LSTM layer, that is, outputting a label sequence. Experiments conducted on corpora of the 2010, 2012 and 2014 i2b2 NLP challenges show that LSTM achieves highest micro-average F1-scores of 85.81% on the 2010 i2b2 medical concept extraction, 92.29% on the 2012 i2b2 clinical event detection, and 94.37% on the 2014 i2b2 de-identification, which is considerably competitive with other state-of-the-art systems. LSTM that requires no hand-crafted feature has great potential on entity recognition from clinical texts. It outperforms traditional machine learning methods that suffer from fussy feature engineering. A possible future direction is how to integrate knowledge
El-Nagar, Ahmad M
2018-01-01
In this study, a novel structure of a recurrent interval type-2 Takagi-Sugeno-Kang (TSK) fuzzy neural network (FNN) is introduced for nonlinear dynamic and time-varying systems identification. It combines the type-2 fuzzy sets (T2FSs) and a recurrent FNN to avoid the data uncertainties. The fuzzy firing strengths in the proposed structure are returned to the network input as internal variables. The interval type-2 fuzzy sets (IT2FSs) is used to describe the antecedent part for each rule while the consequent part is a TSK-type, which is a linear function of the internal variables and the external inputs with interval weights. All the type-2 fuzzy rules for the proposed RIT2TSKFNN are learned on-line based on structure and parameter learning, which are performed using the type-2 fuzzy clustering. The antecedent and consequent parameters of the proposed RIT2TSKFNN are updated based on the Lyapunov function to achieve network stability. The obtained results indicate that our proposed network has a small root mean square error (RMSE) and a small integral of square error (ISE) with a small number of rules and a small computation time compared with other type-2 FNNs. Copyright © 2017 ISA. Published by Elsevier Ltd. All rights reserved.
Liu, Qingshan; Dang, Chuangyin; Huang, Tingwen
2013-02-01
This paper presents a decision-making model described by a recurrent neural network for dynamic portfolio optimization. The portfolio-optimization problem is first converted into a constrained fractional programming problem. Since the objective function in the programming problem is not convex, the traditional optimization techniques are no longer applicable for solving this problem. Fortunately, the objective function in the fractional programming is pseudoconvex on the feasible region. It leads to a one-layer recurrent neural network modeled by means of a discontinuous dynamic system. To ensure the optimal solutions for portfolio optimization, the convergence of the proposed neural network is analyzed and proved. In fact, the neural network guarantees to get the optimal solutions for portfolio-investment advice if some mild conditions are satisfied. A numerical example with simulation results substantiates the effectiveness and illustrates the characteristics of the proposed neural network.
Exponential stability of delayed recurrent neural networks with Markovian jumping parameters
International Nuclear Information System (INIS)
Wang Zidong; Liu Yurong; Yu Li; Liu Xiaohui
2006-01-01
In this Letter, the global exponential stability analysis problem is considered for a class of recurrent neural networks (RNNs) with time delays and Markovian jumping parameters. The jumping parameters considered here are generated from a continuous-time discrete-state homogeneous Markov process, which are governed by a Markov process with discrete and finite state space. The purpose of the problem addressed is to derive some easy-to-test conditions such that the dynamics of the neural network is stochastically exponentially stable in the mean square, independent of the time delay. By employing a new Lyapunov-Krasovskii functional, a linear matrix inequality (LMI) approach is developed to establish the desired sufficient conditions, and therefore the global exponential stability in the mean square for the delayed RNNs can be easily checked by utilizing the numerically efficient Matlab LMI toolbox, and no tuning of parameters is required. A numerical example is exploited to show the usefulness of the derived LMI-based stability conditions
Fault diagnosis of rolling bearings with recurrent neural network-based autoencoders.
Liu, Han; Zhou, Jianzhong; Zheng, Yang; Jiang, Wei; Zhang, Yuncheng
2018-04-19
As the rolling bearings being the key part of rotary machine, its healthy condition is quite important for safety production. Fault diagnosis of rolling bearing has been research focus for the sake of improving the economic efficiency and guaranteeing the operation security. However, the collected signals are mixed with ambient noise during the operation of rotary machine, which brings great challenge to the exact diagnosis results. Using signals collected from multiple sensors can avoid the loss of local information and extract more helpful characteristics. Recurrent Neural Networks (RNN) is a type of artificial neural network which can deal with multiple time sequence data. The capacity of RNN has been proved outstanding for catching time relevance about time sequence data. This paper proposed a novel method for bearing fault diagnosis with RNN in the form of an autoencoder. In this approach, multiple vibration value of the rolling bearings of the next period are predicted from the previous period by means of Gated Recurrent Unit (GRU)-based denoising autoencoder. These GRU-based non-linear predictive denoising autoencoders (GRU-NP-DAEs) are trained with strong generalization ability for each different fault pattern. Then for the given input data, the reconstruction errors between the next period data and the output data generated by different GRU-NP-DAEs are used to detect anomalous conditions and classify fault type. Classic rotating machinery datasets have been employed to testify the effectiveness of the proposed diagnosis method and its preponderance over some state-of-the-art methods. The experiment results indicate that the proposed method achieves satisfactory performance with strong robustness and high classification accuracy. Copyright © 2018 ISA. Published by Elsevier Ltd. All rights reserved.
Predicting recurrent aphthous ulceration using genetic algorithms-optimized neural networks
Directory of Open Access Journals (Sweden)
Najla S Dar-Odeh
2010-05-01
Full Text Available Najla S Dar-Odeh1, Othman M Alsmadi2, Faris Bakri3, Zaer Abu-Hammour2, Asem A Shehabi3, Mahmoud K Al-Omiri1, Shatha M K Abu-Hammad4, Hamzeh Al-Mashni4, Mohammad B Saeed4, Wael Muqbil4, Osama A Abu-Hammad1 1Faculty of Dentistry, 2Faculty of Engineering and Technology, 3Faculty of Medicine, University of Jordan, Amman, Jordan; 4Dental Department, University of Jordan Hospital, Amman, JordanObjective: To construct and optimize a neural network that is capable of predicting the occurrence of recurrent aphthous ulceration (RAU based on a set of appropriate input data.Participants and methods: Artificial neural networks (ANN software employing genetic algorithms to optimize the architecture neural networks was used. Input and output data of 86 participants (predisposing factors and status of the participants with regards to recurrent aphthous ulceration were used to construct and train the neural networks. The optimized neural networks were then tested using untrained data of a further 10 participants.Results: The optimized neural network, which produced the most accurate predictions for the presence or absence of recurrent aphthous ulceration was found to employ: gender, hematological (with or without ferritin and mycological data of the participants, frequency of tooth brushing, and consumption of vegetables and fruits.Conclusions: Factors appearing to be related to recurrent aphthous ulceration and appropriate for use as input data to construct ANNs that predict recurrent aphthous ulceration were found to include the following: gender, hemoglobin, serum vitamin B12, serum ferritin, red cell folate, salivary candidal colony count, frequency of tooth brushing, and the number of fruits or vegetables consumed daily.Keywords: artifical neural networks, recurrent, aphthous ulceration, ulcer
Frequency characteristics of coordinate sequences of linear recurrences over Galois rings
Kamlovskii, O. V.
2013-12-01
We consider some properties of the coordinate sequences of linear recurrences over Galois rings which characterize the possibility of regarding them as pseudo-random sequences. We study the periodicity properties, linear complexity and frequency characteristics of these sequences. Up to now, these parameters have been studied mainly in the case when the linear recurring sequence has maximal possible period. We investigate the coordinate sequences of linear recurrences of not necessarily maximal period. We obtain sharpened and generalized estimates for the number of elements and r-patterns on the cycles and intervals of these sequences.
Frequency characteristics of coordinate sequences of linear recurrences over Galois rings
International Nuclear Information System (INIS)
Certification Research Center, Moscow (Russian Federation))" data-affiliation=" (LLC Certification Research Center, Moscow (Russian Federation))" >Kamlovskii, O V
2013-01-01
We consider some properties of the coordinate sequences of linear recurrences over Galois rings which characterize the possibility of regarding them as pseudo-random sequences. We study the periodicity properties, linear complexity and frequency characteristics of these sequences. Up to now, these parameters have been studied mainly in the case when the linear recurring sequence has maximal possible period. We investigate the coordinate sequences of linear recurrences of not necessarily maximal period. We obtain sharpened and generalized estimates for the number of elements and r-patterns on the cycles and intervals of these sequences
APPLICATION OF NEURAL NETWORK ALGORITHMS FOR BPM LINEARIZATION
Energy Technology Data Exchange (ETDEWEB)
Musson, John C. [JLAB; Seaton, Chad [JLAB; Spata, Mike F. [JLAB; Yan, Jianxun [JLAB
2012-11-01
Stripline BPM sensors contain inherent non-linearities, as a result of field distortions from the pickup elements. Many methods have been devised to facilitate corrections, often employing polynomial fitting. The cost of computation makes real-time correction difficult, particulalry when integer math is utilized. The application of neural-network technology, particularly the multi-layer perceptron algorithm, is proposed as an efficient alternative for electrode linearization. A process of supervised learning is initially used to determine the weighting coefficients, which are subsequently applied to the incoming electrode data. A non-linear layer, known as an activation layer, is responsible for the removal of saturation effects. Implementation of a perceptron in an FPGA-based software-defined radio (SDR) is presented, along with performance comparisons. In addition, efficient calculation of the sigmoidal activation function via the CORDIC algorithm is presented.
International Nuclear Information System (INIS)
Wang Shen-Quan; Feng Jian; Zhao Qing
2012-01-01
In this paper, the problem of delay-distribution-dependent stability is investigated for continuous-time recurrent neural networks (CRNNs) with stochastic delay. Different from the common assumptions on time delays, it is assumed that the probability distribution of the delay taking values in some intervals is known a priori. By making full use of the information concerning the probability distribution of the delay and by using a tighter bounding technique (the reciprocally convex combination method), less conservative asymptotic mean-square stable sufficient conditions are derived in terms of linear matrix inequalities (LMIs). Two numerical examples show that our results are better than the existing ones. (general)
Ads' click-through rates predicting based on gated recurrent unit neural networks
Chen, Qiaohong; Guo, Zixuan; Dong, Wen; Jin, Lingzi
2018-05-01
In order to improve the effect of online advertising and to increase the revenue of advertising, the gated recurrent unit neural networks(GRU) model is used as the ads' click through rates(CTR) predicting. Combined with the characteristics of gated unit structure and the unique of time sequence in data, using BPTT algorithm to train the model. Furthermore, by optimizing the step length algorithm of the gated unit recurrent neural networks, making the model reach optimal point better and faster in less iterative rounds. The experiment results show that the model based on the gated recurrent unit neural networks and its optimization of step length algorithm has the better effect on the ads' CTR predicting, which helps advertisers, media and audience achieve a win-win and mutually beneficial situation in Three-Side Game.
Multistability and instability analysis of recurrent neural networks with time-varying delays.
Zhang, Fanghai; Zeng, Zhigang
2018-01-01
This paper provides new theoretical results on the multistability and instability analysis of recurrent neural networks with time-varying delays. It is shown that such n-neuronal recurrent neural networks have exactly [Formula: see text] equilibria, [Formula: see text] of which are locally exponentially stable and the others are unstable, where k 0 is a nonnegative integer such that k 0 ≤n. By using the combination method of two different divisions, recurrent neural networks can possess more dynamic properties. This method improves and extends the existing results in the literature. Finally, one numerical example is provided to show the superiority and effectiveness of the presented results. Copyright © 2017 Elsevier Ltd. All rights reserved.
Financial Time Series Prediction Using Elman Recurrent Random Neural Networks
Directory of Open Access Journals (Sweden)
Jie Wang
2016-01-01
(ERNN, the empirical results show that the proposed neural network displays the best performance among these neural networks in financial time series forecasting. Further, the empirical research is performed in testing the predictive effects of SSE, TWSE, KOSPI, and Nikkei225 with the established model, and the corresponding statistical comparisons of the above market indices are also exhibited. The experimental results show that this approach gives good performance in predicting the values from the stock market indices.
Recurrent neural network based hybrid model for reconstructing gene regulatory network.
Raza, Khalid; Alam, Mansaf
2016-10-01
One of the exciting problems in systems biology research is to decipher how genome controls the development of complex biological system. The gene regulatory networks (GRNs) help in the identification of regulatory interactions between genes and offer fruitful information related to functional role of individual gene in a cellular system. Discovering GRNs lead to a wide range of applications, including identification of disease related pathways providing novel tentative drug targets, helps to predict disease response, and also assists in diagnosing various diseases including cancer. Reconstruction of GRNs from available biological data is still an open problem. This paper proposes a recurrent neural network (RNN) based model of GRN, hybridized with generalized extended Kalman filter for weight update in backpropagation through time training algorithm. The RNN is a complex neural network that gives a better settlement between biological closeness and mathematical flexibility to model GRN; and is also able to capture complex, non-linear and dynamic relationships among variables. Gene expression data are inherently noisy and Kalman filter performs well for estimation problem even in noisy data. Hence, we applied non-linear version of Kalman filter, known as generalized extended Kalman filter, for weight update during RNN training. The developed model has been tested on four benchmark networks such as DNA SOS repair network, IRMA network, and two synthetic networks from DREAM Challenge. We performed a comparison of our results with other state-of-the-art techniques which shows superiority of our proposed model. Further, 5% Gaussian noise has been induced in the dataset and result of the proposed model shows negligible effect of noise on results, demonstrating the noise tolerance capability of the model. Copyright © 2016 Elsevier Ltd. All rights reserved.
DEFF Research Database (Denmark)
Grinke, Eduard; Tetzlaff, Christian; Wörgötter, Florentin
2015-01-01
correlation-based learning with synaptic scaling is applied to adequately change the connections of the network. By doing so, we can effectively exploit neural dynamics (i.e., hysteresis effects and single attractors) in the network to generate different turning angles with short-term memory for a walking...... dynamics, plasticity, sensory feedback, and biomechanics. Generating such versatile and adaptive behaviors for a many degrees-of-freedom (DOFs) walking robot is a challenging task. Thus, in this study, we present a bio-inspired approach to solve this task. Specifically, the approach combines neural...... mechanisms with plasticity, exteroceptive sensory feedback, and biomechanics. The neural mechanisms consist of adaptive neural sensory processing and modular neural locomotion control. The sensory processing is based on a small recurrent neural network consisting of two fully connected neurons. Online...
A novel nonlinear adaptive filter using a pipelined second-order Volterra recurrent neural network.
Zhao, Haiquan; Zhang, Jiashu
2009-12-01
To enhance the performance and overcome the heavy computational complexity of recurrent neural networks (RNN), a novel nonlinear adaptive filter based on a pipelined second-order Volterra recurrent neural network (PSOVRNN) is proposed in this paper. A modified real-time recurrent learning (RTRL) algorithm of the proposed filter is derived in much more detail. The PSOVRNN comprises of a number of simple small-scale second-order Volterra recurrent neural network (SOVRNN) modules. In contrast to the standard RNN, these modules of a PSOVRNN can be performed simultaneously in a pipelined parallelism fashion, which can lead to a significant improvement in its total computational efficiency. Moreover, since each module of the PSOVRNN is a SOVRNN in which nonlinearity is introduced by the recursive second-order Volterra (RSOV) expansion, its performance can be further improved. Computer simulations have demonstrated that the PSOVRNN performs better than the pipelined recurrent neural network (PRNN) and RNN for nonlinear colored signals prediction and nonlinear channel equalization. However, the superiority of the PSOVRNN over the PRNN is at the cost of increasing computational complexity due to the introduced nonlinear expansion of each module.
Lin, Yang-Yin; Chang, Jyh-Yeong; Lin, Chin-Teng
2013-02-01
This paper presents a novel recurrent fuzzy neural network, called an interactively recurrent self-evolving fuzzy neural network (IRSFNN), for prediction and identification of dynamic systems. The recurrent structure in an IRSFNN is formed as an external loops and internal feedback by feeding the rule firing strength of each rule to others rules and itself. The consequent part in the IRSFNN is composed of a Takagi-Sugeno-Kang (TSK) or functional-link-based type. The proposed IRSFNN employs a functional link neural network (FLNN) to the consequent part of fuzzy rules for promoting the mapping ability. Unlike a TSK-type fuzzy neural network, the FLNN in the consequent part is a nonlinear function of input variables. An IRSFNNs learning starts with an empty rule base and all of the rules are generated and learned online through a simultaneous structure and parameter learning. An on-line clustering algorithm is effective in generating fuzzy rules. The consequent update parameters are derived by a variable-dimensional Kalman filter algorithm. The premise and recurrent parameters are learned through a gradient descent algorithm. We test the IRSFNN for the prediction and identification of dynamic plants and compare it to other well-known recurrent FNNs. The proposed model obtains enhanced performance results.
International Nuclear Information System (INIS)
Liang Jinling; Cao Jinde
2003-01-01
In this Letter, the problems of boundedness and stability for a general class of non-autonomous recurrent neural networks with variable coefficients and time-varying delays are analyzed via employing Young inequality technique and Lyapunov method. Some simple sufficient conditions are given for boundedness and stability of the solutions for the recurrent neural networks. These results generalize and improve the previous works, and they are easy to check and apply in practice. Two illustrative examples and their numerical simulations are also given to demonstrate the effectiveness of the proposed results
Hybrid Spectral Unmixing: Using Artificial Neural Networks for Linear/Non-Linear Switching
Directory of Open Access Journals (Sweden)
Asmau M. Ahmed
2017-07-01
Full Text Available Spectral unmixing is a key process in identifying spectral signature of materials and quantifying their spatial distribution over an image. The linear model is expected to provide acceptable results when two assumptions are satisfied: (1 The mixing process should occur at macroscopic level and (2 Photons must interact with single material before reaching the sensor. However, these assumptions do not always hold and more complex nonlinear models are required. This study proposes a new hybrid method for switching between linear and nonlinear spectral unmixing of hyperspectral data based on artificial neural networks. The neural networks was trained with parameters within a window of the pixel under consideration. These parameters are computed to represent the diversity of the neighboring pixels and are based on the Spectral Angular Distance, Covariance and a non linearity parameter. The endmembers were extracted using Vertex Component Analysis while the abundances were estimated using the method identified by the neural networks (Vertex Component Analysis, Fully Constraint Least Square Method, Polynomial Post Nonlinear Mixing Model or Generalized Bilinear Model. Results show that the hybrid method performs better than each of the individual techniques with high overall accuracy, while the abundance estimation error is significantly lower than that obtained using the individual methods. Experiments on both synthetic dataset and real hyperspectral images demonstrated that the proposed hybrid switch method is efficient for solving spectral unmixing of hyperspectral images as compared to individual algorithms.
Training the Recurrent neural network by the Fuzzy Min-Max algorithm for fault prediction
International Nuclear Information System (INIS)
Zemouri, Ryad; Racoceanu, Daniel; Zerhouni, Noureddine; Minca, Eugenia; Filip, Florin
2009-01-01
In this paper, we present a training technique of a Recurrent Radial Basis Function neural network for fault prediction. We use the Fuzzy Min-Max technique to initialize the k-center of the RRBF neural network. The k-means algorithm is then applied to calculate the centers that minimize the mean square error of the prediction task. The performances of the k-means algorithm are then boosted by the Fuzzy Min-Max technique.
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Francisco Javier Ordóñez; Daniel Roggen
2016-01-01
Human activity recognition (HAR) tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we pro...
Bi-directional LSTM Recurrent Neural Network for Chinese Word Segmentation
Yao, Yushi; Huang, Zheng
2016-01-01
Recurrent neural network(RNN) has been broadly applied to natural language processing(NLP) problems. This kind of neural network is designed for modeling sequential data and has been testified to be quite efficient in sequential tagging tasks. In this paper, we propose to use bi-directional RNN with long short-term memory(LSTM) units for Chinese word segmentation, which is a crucial preprocess task for modeling Chinese sentences and articles. Classical methods focus on designing and combining...
International Nuclear Information System (INIS)
Lou, X.; Cui, B.
2008-01-01
In this paper we consider the problem of exponential stability for recurrent neural networks with multiple time varying delays and reaction-diffusion terms. The activation functions are supposed to be bounded and globally Lipschitz continuous. By means of Lyapunov functional, sufficient conditions are derived, which guarantee global exponential stability of the delayed neural network. Finally, a numerical example is given to show the correctness of our analysis. (author)
ReSeg: A Recurrent Neural Network-Based Model for Semantic Segmentation
Visin, Francesco; Ciccone, Marco; Romero, Adriana; Kastner, Kyle; Cho, Kyunghyun; Bengio, Yoshua; Matteucci, Matteo; Courville, Aaron
2015-01-01
We propose a structured prediction architecture, which exploits the local generic features extracted by Convolutional Neural Networks and the capacity of Recurrent Neural Networks (RNN) to retrieve distant dependencies. The proposed architecture, called ReSeg, is based on the recently introduced ReNet model for image classification. We modify and extend it to perform the more challenging task of semantic segmentation. Each ReNet layer is composed of four RNN that sweep the image horizontally ...
Reduced-Order Modeling for Flutter/LCO Using Recurrent Artificial Neural Network
Yao, Weigang; Liou, Meng-Sing
2012-01-01
The present study demonstrates the efficacy of a recurrent artificial neural network to provide a high fidelity time-dependent nonlinear reduced-order model (ROM) for flutter/limit-cycle oscillation (LCO) modeling. An artificial neural network is a relatively straightforward nonlinear method for modeling an input-output relationship from a set of known data, for which we use the radial basis function (RBF) with its parameters determined through a training process. The resulting RBF neural network, however, is only static and is not yet adequate for an application to problems of dynamic nature. The recurrent neural network method [1] is applied to construct a reduced order model resulting from a series of high-fidelity time-dependent data of aero-elastic simulations. Once the RBF neural network ROM is constructed properly, an accurate approximate solution can be obtained at a fraction of the cost of a full-order computation. The method derived during the study has been validated for predicting nonlinear aerodynamic forces in transonic flow and is capable of accurate flutter/LCO simulations. The obtained results indicate that the present recurrent RBF neural network is accurate and efficient for nonlinear aero-elastic system analysis
Stimulus-dependent suppression of chaos in recurrent neural networks
International Nuclear Information System (INIS)
Rajan, Kanaka; Abbott, L. F.; Sompolinsky, Haim
2010-01-01
Neuronal activity arises from an interaction between ongoing firing generated spontaneously by neural circuits and responses driven by external stimuli. Using mean-field analysis, we ask how a neural network that intrinsically generates chaotic patterns of activity can remain sensitive to extrinsic input. We find that inputs not only drive network responses, but they also actively suppress ongoing activity, ultimately leading to a phase transition in which chaos is completely eliminated. The critical input intensity at the phase transition is a nonmonotonic function of stimulus frequency, revealing a 'resonant' frequency at which the input is most effective at suppressing chaos even though the power spectrum of the spontaneous activity peaks at zero and falls exponentially. A prediction of our analysis is that the variance of neural responses should be most strongly suppressed at frequencies matching the range over which many sensory systems operate.
Tuning Recurrent Neural Networks for Recognizing Handwritten Arabic Words
Qaralleh, Esam; Abandah, Gheith; Jamour, Fuad Tarek
2013-01-01
and sizes of the hidden layers. Large sizes are slow and small sizes are generally not accurate. Tuning the neural network size is a hard task because the design space is often large and training is often a long process. We use design of experiments
Recurrent Artificial Neural Networks and Finite State Natural Language Processing.
Moisl, Hermann
It is argued that pessimistic assessments of the adequacy of artificial neural networks (ANNs) for natural language processing (NLP) on the grounds that they have a finite state architecture are unjustified, and that their adequacy in this regard is an empirical issue. First, arguments that counter standard objections to finite state NLP on the…
Homeostatic scaling of excitability in recurrent neural networks.
Remme, M.W.H.; Wadman, W.J.
2012-01-01
Neurons adjust their intrinsic excitability when experiencing a persistent change in synaptic drive. This process can prevent neural activity from moving into either a quiescent state or a saturated state in the face of ongoing plasticity, and is thought to promote stability of the network in which
Individual Identification Using Functional Brain Fingerprint Detected by Recurrent Neural Network.
Chen, Shiyang; Hu, Xiaoping P
2018-03-20
Individual identification based on brain function has gained traction in literature. Investigating individual differences in brain function can provide additional insights into the brain. In this work, we introduce a recurrent neural network based model for identifying individuals based on only a short segment of resting state functional MRI data. In addition, we demonstrate how the global signal and differences in atlases affect the individual identifiability. Furthermore, we investigate neural network features that exhibit the uniqueness of each individual. The results indicate that our model is able to identify individuals based on neural features and provides additional information regarding brain dynamics.
Qin, Sitian; Yang, Xiudong; Xue, Xiaoping; Song, Jiahui
2017-10-01
Pseudoconvex optimization problem, as an important nonconvex optimization problem, plays an important role in scientific and engineering applications. In this paper, a recurrent one-layer neural network is proposed for solving the pseudoconvex optimization problem with equality and inequality constraints. It is proved that from any initial state, the state of the proposed neural network reaches the feasible region in finite time and stays there thereafter. It is also proved that the state of the proposed neural network is convergent to an optimal solution of the related problem. Compared with the related existing recurrent neural networks for the pseudoconvex optimization problems, the proposed neural network in this paper does not need the penalty parameters and has a better convergence. Meanwhile, the proposed neural network is used to solve three nonsmooth optimization problems, and we make some detailed comparisons with the known related conclusions. In the end, some numerical examples are provided to illustrate the effectiveness of the performance of the proposed neural network.
Folk music style modelling by recurrent neural networks with long short term memory units
Sturm, Bob; Santos, João Felipe; Korshunova, Iryna
2015-01-01
We demonstrate two generative models created by training a recurrent neural network (RNN) with three hidden layers of long short-term memory (LSTM) units. This extends past work in numerous directions, including training deeper models with nearly 24,000 high-level transcriptions of folk tunes. We discuss our on-going work.
Recurrent Neural Network For Forecasting Time Series With Long Memory Pattern
Walid; Alamsyah
2017-04-01
Recurrent Neural Network as one of the hybrid models are often used to predict and estimate the issues related to electricity, can be used to describe the cause of the swelling of electrical load which experienced by PLN. In this research will be developed RNN forecasting procedures at the time series with long memory patterns. Considering the application is the national electrical load which of course has a different trend with the condition of the electrical load in any country. This research produces the algorithm of time series forecasting which has long memory pattern using E-RNN after this referred to the algorithm of integrated fractional recurrent neural networks (FIRNN).The prediction results of long memory time series using models Fractional Integrated Recurrent Neural Network (FIRNN) showed that the model with the selection of data difference in the range of [-1,1] and the model of Fractional Integrated Recurrent Neural Network (FIRNN) (24,6,1) provides the smallest MSE value, which is 0.00149684.
Encoding of phonology in a recurrent neural model of grounded speech
Alishahi, Afra; Barking, Marie; Chrupala, Grzegorz; Levy, Roger; Specia, Lucia
2017-01-01
We study the representation and encoding of phonemes in a recurrent neural network model of grounded speech. We use a model which processes images and their spoken descriptions, and projects the visual and auditory representations into the same semantic space. We perform a number of analyses on how
Direction-of-change forecasting using a volatility-based recurrent neural network
Bekiros, S.D.; Georgoutsos, D.A.
2008-01-01
This paper investigates the profitability of a trading strategy, based on recurrent neural networks, that attempts to predict the direction-of-change of the market in the case of the NASDAQ composite index. The sample extends over the period 8 February 1971 to 7 April 1998, while the sub-period 8
Global stability of discrete-time recurrent neural networks with impulse effects
International Nuclear Information System (INIS)
Zhou, L; Li, C; Wan, J
2008-01-01
This paper formulates and studies a class of discrete-time recurrent neural networks with impulse effects. A stability criterion, which characterizes the effects of impulse and stability property of the corresponding impulse-free networks on the stability of the impulsive networks in an aggregate form, is established. Two simplified and numerically tractable criteria are also provided
A one-layer recurrent neural network for constrained nonsmooth optimization.
Liu, Qingshan; Wang, Jun
2011-10-01
This paper presents a novel one-layer recurrent neural network modeled by means of a differential inclusion for solving nonsmooth optimization problems, in which the number of neurons in the proposed neural network is the same as the number of decision variables of optimization problems. Compared with existing neural networks for nonsmooth optimization problems, the global convexity condition on the objective functions and constraints is relaxed, which allows the objective functions and constraints to be nonconvex. It is proven that the state variables of the proposed neural network are convergent to optimal solutions if a single design parameter in the model is larger than a derived lower bound. Numerical examples with simulation results substantiate the effectiveness and illustrate the characteristics of the proposed neural network.
A one-layer recurrent neural network for constrained nonconvex optimization.
Li, Guocheng; Yan, Zheng; Wang, Jun
2015-01-01
In this paper, a one-layer recurrent neural network is proposed for solving nonconvex optimization problems subject to general inequality constraints, designed based on an exact penalty function method. It is proved herein that any neuron state of the proposed neural network is convergent to the feasible region in finite time and stays there thereafter, provided that the penalty parameter is sufficiently large. The lower bounds of the penalty parameter and convergence time are also estimated. In addition, any neural state of the proposed neural network is convergent to its equilibrium point set which satisfies the Karush-Kuhn-Tucker conditions of the optimization problem. Moreover, the equilibrium point set is equivalent to the optimal solution to the nonconvex optimization problem if the objective function and constraints satisfy given conditions. Four numerical examples are provided to illustrate the performances of the proposed neural network.
A one-layer recurrent neural network for constrained nonsmooth invex optimization.
Li, Guocheng; Yan, Zheng; Wang, Jun
2014-02-01
Invexity is an important notion in nonconvex optimization. In this paper, a one-layer recurrent neural network is proposed for solving constrained nonsmooth invex optimization problems, designed based on an exact penalty function method. It is proved herein that any state of the proposed neural network is globally convergent to the optimal solution set of constrained invex optimization problems, with a sufficiently large penalty parameter. In addition, any neural state is globally convergent to the unique optimal solution, provided that the objective function and constraint functions are pseudoconvex. Moreover, any neural state is globally convergent to the feasible region in finite time and stays there thereafter. The lower bounds of the penalty parameter and convergence time are also estimated. Two numerical examples are provided to illustrate the performances of the proposed neural network. Copyright © 2013 Elsevier Ltd. All rights reserved.
Modeling long-term human activeness using recurrent neural networks for biometric data.
Kim, Zae Myung; Oh, Hyungrai; Kim, Han-Gyu; Lim, Chae-Gyun; Oh, Kyo-Joong; Choi, Ho-Jin
2017-05-18
With the invention of fitness trackers, it has been possible to continuously monitor a user's biometric data such as heart rates, number of footsteps taken, and amount of calories burned. This paper names the time series of these three types of biometric data, the user's "activeness", and investigates the feasibility in modeling and predicting the long-term activeness of the user. The dataset used in this study consisted of several months of biometric time-series data gathered by seven users independently. Four recurrent neural network (RNN) architectures-as well as a deep neural network and a simple regression model-were proposed to investigate the performance on predicting the activeness of the user under various length-related hyper-parameter settings. In addition, the learned model was tested to predict the time period when the user's activeness falls below a certain threshold. A preliminary experimental result shows that each type of activeness data exhibited a short-term autocorrelation; and among the three types of data, the consumed calories and the number of footsteps were positively correlated, while the heart rate data showed almost no correlation with neither of them. It is probably due to this characteristic of the dataset that although the RNN models produced the best results on modeling the user's activeness, the difference was marginal; and other baseline models, especially the linear regression model, performed quite admirably as well. Further experimental results show that it is feasible to predict a user's future activeness with precision, for example, a trained RNN model could predict-with the precision of 84%-when the user would be less active within the next hour given the latest 15 min of his activeness data. This paper defines and investigates the notion of a user's "activeness", and shows that forecasting the long-term activeness of the user is indeed possible. Such information can be utilized by a health-related application to proactively
Hysteretic recurrent neural networks: a tool for modeling hysteretic materials and systems
International Nuclear Information System (INIS)
Veeramani, Arun S; Crews, John H; Buckner, Gregory D
2009-01-01
This paper introduces a novel recurrent neural network, the hysteretic recurrent neural network (HRNN), that is ideally suited to modeling hysteretic materials and systems. This network incorporates a hysteretic neuron consisting of conjoined sigmoid activation functions. Although similar hysteretic neurons have been explored previously, the HRNN is unique in its utilization of simple recurrence to 'self-select' relevant activation functions. Furthermore, training is facilitated by placing the network weights on the output side, allowing standard backpropagation of error training algorithms to be used. We present two- and three-phase versions of the HRNN for modeling hysteretic materials with distinct phases. These models are experimentally validated using data collected from shape memory alloys and ferromagnetic materials. The results demonstrate the HRNN's ability to accurately generalize hysteretic behavior with a relatively small number of neurons. Additional benefits lie in the network's ability to identify statistical information concerning the macroscopic material by analyzing the weights of the individual neurons
Spatiotemporal Dynamics and Reliable Computations in Recurrent Spiking Neural Networks
Pyle, Ryan; Rosenbaum, Robert
2017-01-01
Randomly connected networks of excitatory and inhibitory spiking neurons provide a parsimonious model of neural variability, but are notoriously unreliable for performing computations. We show that this difficulty is overcome by incorporating the well-documented dependence of connection probability on distance. Spatially extended spiking networks exhibit symmetry-breaking bifurcations and generate spatiotemporal patterns that can be trained to perform dynamical computations under a reservoir computing framework.
Spatiotemporal Dynamics and Reliable Computations in Recurrent Spiking Neural Networks.
Pyle, Ryan; Rosenbaum, Robert
2017-01-06
Randomly connected networks of excitatory and inhibitory spiking neurons provide a parsimonious model of neural variability, but are notoriously unreliable for performing computations. We show that this difficulty is overcome by incorporating the well-documented dependence of connection probability on distance. Spatially extended spiking networks exhibit symmetry-breaking bifurcations and generate spatiotemporal patterns that can be trained to perform dynamical computations under a reservoir computing framework.
On Sequences of Numbers and Polynomials Defined by Linear Recurrence Relations of Order 2
Directory of Open Access Journals (Sweden)
Tian-Xiao He
2009-01-01
Full Text Available Here we present a new method to construct the explicit formula of a sequence of numbers and polynomials generated by a linear recurrence relation of order 2. The applications of the method to the Fibonacci and Lucas numbers, Chebyshev polynomials, the generalized Gegenbauer-Humbert polynomials are also discussed. The derived idea provides a general method to construct identities of number or polynomial sequences defined by linear recurrence relations. The applications using the method to solve some algebraic and ordinary differential equations are presented.
Firing rate dynamics in recurrent spiking neural networks with intrinsic and network heterogeneity.
Ly, Cheng
2015-12-01
Heterogeneity of neural attributes has recently gained a lot of attention and is increasing recognized as a crucial feature in neural processing. Despite its importance, this physiological feature has traditionally been neglected in theoretical studies of cortical neural networks. Thus, there is still a lot unknown about the consequences of cellular and circuit heterogeneity in spiking neural networks. In particular, combining network or synaptic heterogeneity and intrinsic heterogeneity has yet to be considered systematically despite the fact that both are known to exist and likely have significant roles in neural network dynamics. In a canonical recurrent spiking neural network model, we study how these two forms of heterogeneity lead to different distributions of excitatory firing rates. To analytically characterize how these types of heterogeneities affect the network, we employ a dimension reduction method that relies on a combination of Monte Carlo simulations and probability density function equations. We find that the relationship between intrinsic and network heterogeneity has a strong effect on the overall level of heterogeneity of the firing rates. Specifically, this relationship can lead to amplification or attenuation of firing rate heterogeneity, and these effects depend on whether the recurrent network is firing asynchronously or rhythmically firing. These observations are captured with the aforementioned reduction method, and furthermore simpler analytic descriptions based on this dimension reduction method are developed. The final analytic descriptions provide compact and descriptive formulas for how the relationship between intrinsic and network heterogeneity determines the firing rate heterogeneity dynamics in various settings.
Land Cover Classification via Multitemporal Spatial Data by Deep Recurrent Neural Networks
Ienco, Dino; Gaetano, Raffaele; Dupaquier, Claire; Maurel, Pierre
2017-10-01
Nowadays, modern earth observation programs produce huge volumes of satellite images time series (SITS) that can be useful to monitor geographical areas through time. How to efficiently analyze such kind of information is still an open question in the remote sensing field. Recently, deep learning methods proved suitable to deal with remote sensing data mainly for scene classification (i.e. Convolutional Neural Networks - CNNs - on single images) while only very few studies exist involving temporal deep learning approaches (i.e Recurrent Neural Networks - RNNs) to deal with remote sensing time series. In this letter we evaluate the ability of Recurrent Neural Networks, in particular the Long-Short Term Memory (LSTM) model, to perform land cover classification considering multi-temporal spatial data derived from a time series of satellite images. We carried out experiments on two different datasets considering both pixel-based and object-based classification. The obtained results show that Recurrent Neural Networks are competitive compared to state-of-the-art classifiers, and may outperform classical approaches in presence of low represented and/or highly mixed classes. We also show that using the alternative feature representation generated by LSTM can improve the performances of standard classifiers.
Artificial Neural Network versus Linear Models Forecasting Doha Stock Market
Yousif, Adil; Elfaki, Faiz
2017-12-01
The purpose of this study is to determine the instability of Doha stock market and develop forecasting models. Linear time series models are used and compared with a nonlinear Artificial Neural Network (ANN) namely Multilayer Perceptron (MLP) Technique. It aims to establish the best useful model based on daily and monthly data which are collected from Qatar exchange for the period starting from January 2007 to January 2015. Proposed models are for the general index of Qatar stock exchange and also for the usages in other several sectors. With the help of these models, Doha stock market index and other various sectors were predicted. The study was conducted by using various time series techniques to study and analyze data trend in producing appropriate results. After applying several models, such as: Quadratic trend model, double exponential smoothing model, and ARIMA, it was concluded that ARIMA (2,2) was the most suitable linear model for the daily general index. However, ANN model was found to be more accurate than time series models.
Kordmahalleh, Mina Moradi; Sefidmazgi, Mohammad Gorji; Harrison, Scott H; Homaifar, Abdollah
2017-01-01
The modeling of genetic interactions within a cell is crucial for a basic understanding of physiology and for applied areas such as drug design. Interactions in gene regulatory networks (GRNs) include effects of transcription factors, repressors, small metabolites, and microRNA species. In addition, the effects of regulatory interactions are not always simultaneous, but can occur after a finite time delay, or as a combined outcome of simultaneous and time delayed interactions. Powerful biotechnologies have been rapidly and successfully measuring levels of genetic expression to illuminate different states of biological systems. This has led to an ensuing challenge to improve the identification of specific regulatory mechanisms through regulatory network reconstructions. Solutions to this challenge will ultimately help to spur forward efforts based on the usage of regulatory network reconstructions in systems biology applications. We have developed a hierarchical recurrent neural network (HRNN) that identifies time-delayed gene interactions using time-course data. A customized genetic algorithm (GA) was used to optimize hierarchical connectivity of regulatory genes and a target gene. The proposed design provides a non-fully connected network with the flexibility of using recurrent connections inside the network. These features and the non-linearity of the HRNN facilitate the process of identifying temporal patterns of a GRN. Our HRNN method was implemented with the Python language. It was first evaluated on simulated data representing linear and nonlinear time-delayed gene-gene interaction models across a range of network sizes and variances of noise. We then further demonstrated the capability of our method in reconstructing GRNs of the Saccharomyces cerevisiae synthetic network for in vivo benchmarking of reverse-engineering and modeling approaches (IRMA). We compared the performance of our method to TD-ARACNE, HCC-CLINDE, TSNI and ebdbNet across different network
Simultaneous multichannel signal transfers via chaos in a recurrent neural network.
Soma, Ken-ichiro; Mori, Ryota; Sato, Ryuichi; Furumai, Noriyuki; Nara, Shigetoshi
2015-05-01
We propose neural network model that demonstrates the phenomenon of signal transfer between separated neuron groups via other chaotic neurons that show no apparent correlations with the input signal. The model is a recurrent neural network in which it is supposed that synchronous behavior between small groups of input and output neurons has been learned as fragments of high-dimensional memory patterns, and depletion of neural connections results in chaotic wandering dynamics. Computer experiments show that when a strong oscillatory signal is applied to an input group in the chaotic regime, the signal is successfully transferred to the corresponding output group, although no correlation is observed between the input signal and the intermediary neurons. Signal transfer is also observed when multiple signals are applied simultaneously to separate input groups belonging to different memory attractors. In this sense simultaneous multichannel communications are realized, and the chaotic neural dynamics acts as a signal transfer medium in which the signal appears to be hidden.
A non-penalty recurrent neural network for solving a class of constrained optimization problems.
Hosseini, Alireza
2016-01-01
In this paper, we explain a methodology to analyze convergence of some differential inclusion-based neural networks for solving nonsmooth optimization problems. For a general differential inclusion, we show that if its right hand-side set valued map satisfies some conditions, then solution trajectory of the differential inclusion converges to optimal solution set of its corresponding in optimization problem. Based on the obtained methodology, we introduce a new recurrent neural network for solving nonsmooth optimization problems. Objective function does not need to be convex on R(n) nor does the new neural network model require any penalty parameter. We compare our new method with some penalty-based and non-penalty based models. Moreover for differentiable cases, we implement circuit diagram of the new neural network. Copyright © 2015 Elsevier Ltd. All rights reserved.
A Novel Recurrent Neural Network for Manipulator Control With Improved Noise Tolerance.
Li, Shuai; Wang, Huanqing; Rafique, Muhammad Usman
2017-04-12
In this paper, we propose a novel recurrent neural network to resolve the redundancy of manipulators for efficient kinematic control in the presence of noises in a polynomial type. Leveraging the high-order derivative properties of polynomial noises, a deliberately devised neural network is proposed to eliminate the impact of noises and recover the accurate tracking of desired trajectories in workspace. Rigorous analysis shows that the proposed neural law stabilizes the system dynamics and the position tracking error converges to zero in the presence of noises. Extensive simulations verify the theoretical results. Numerical comparisons show that existing dual neural solutions lose stability when exposed to large constant noises or time-varying noises. In contrast, the proposed approach works well and has a low tracking error comparable to noise-free situations.
Multi-stability and almost periodic solutions of a class of recurrent neural networks
International Nuclear Information System (INIS)
Liu Yiguang; You Zhisheng
2007-01-01
This paper studies multi-stability, existence of almost periodic solutions of a class of recurrent neural networks with bounded activation functions. After introducing a sufficient condition insuring multi-stability, many criteria guaranteeing existence of almost periodic solutions are derived using Mawhin's coincidence degree theory. All the criteria are constructed without assuming the activation functions are smooth, monotonic or Lipschitz continuous, and that the networks contains periodic variables (such as periodic coefficients, periodic inputs or periodic activation functions), so all criteria can be easily extended to fit many concrete forms of neural networks such as Hopfield neural networks, or cellular neural networks, etc. Finally, all kinds of simulations are employed to illustrate the criteria
Linearly Recurrent Circle Map Subshifts and an Application to Schrödinger Operators
Adamczewski, B
2001-01-01
We discuss circle map sequences and subshifts generated by them. We give a characterization of those sequences among them which are linearly recurrent. As an application we deduce zero-measure spectrum for a class of discrete one-dimensional Schrödinger operators with potentials generated by circle maps.
International Nuclear Information System (INIS)
Huang Yu-Jiao; Hu Hai-Gen
2015-01-01
In this paper, the multistability issue is discussed for delayed complex-valued recurrent neural networks with discontinuous real-imaginary-type activation functions. Based on a fixed theorem and stability definition, sufficient criteria are established for the existence and stability of multiple equilibria of complex-valued recurrent neural networks. The number of stable equilibria is larger than that of real-valued recurrent neural networks, which can be used to achieve high-capacity associative memories. One numerical example is provided to show the effectiveness and superiority of the presented results. (paper)
Sharma, Richa; Kumar, Vikas; Gaur, Prerna; Mittal, A P
2016-05-01
Being complex, non-linear and coupled system, the robotic manipulator cannot be effectively controlled using classical proportional-integral-derivative (PID) controller. To enhance the effectiveness of the conventional PID controller for the nonlinear and uncertain systems, gains of the PID controller should be conservatively tuned and should adapt to the process parameter variations. In this work, a mix locally recurrent neural network (MLRNN) architecture is investigated to mimic a conventional PID controller which consists of at most three hidden nodes which act as proportional, integral and derivative node. The gains of the mix locally recurrent neural network based PID (MLRNNPID) controller scheme are initialized with a newly developed cuckoo search algorithm (CSA) based optimization method rather than assuming randomly. A sequential learning based least square algorithm is then investigated for the on-line adaptation of the gains of MLRNNPID controller. The performance of the proposed controller scheme is tested against the plant parameters uncertainties and external disturbances for both links of the two link robotic manipulator with variable payload (TL-RMWVP). The stability of the proposed controller is analyzed using Lyapunov stability criteria. A performance comparison is carried out among MLRNNPID controller, CSA optimized NNPID (OPTNNPID) controller and CSA optimized conventional PID (OPTPID) controller in order to establish the effectiveness of the MLRNNPID controller. Copyright © 2016 ISA. Published by Elsevier Ltd. All rights reserved.
Chen, Guiling; Li, Dingshi; Shi, Lin; van Gaans, Onno; Verduyn Lunel, Sjoerd
2018-03-01
We present new conditions for asymptotic stability and exponential stability of a class of stochastic recurrent neural networks with discrete and distributed time varying delays. Our approach is based on the method using fixed point theory, which do not resort to any Liapunov function or Liapunov functional. Our results neither require the boundedness, monotonicity and differentiability of the activation functions nor differentiability of the time varying delays. In particular, a class of neural networks without stochastic perturbations is also considered. Examples are given to illustrate our main results.
International Nuclear Information System (INIS)
Han, Seong Ik; Jeong, Chan Se; Yang, Soon Yong
2012-01-01
A robust positioning control scheme has been developed using friction parameter observer and recurrent fuzzy neural networks based on the sliding mode control. As a dynamic friction model, the LuGre model is adopted for handling friction compensation because it has been known to capture sufficiently the properties of a nonlinear dynamic friction. A developed friction parameter observer has a simple structure and also well estimates friction parameters of the LuGre friction model. In addition, an approximation method for the system uncertainty is developed using recurrent fuzzy neural networks technology to improve the precision positioning degree. Some simulation and experiment provide the verification on the performance of a proposed robust control scheme
CloudScan - A Configuration-Free Invoice Analysis System Using Recurrent Neural Networks
DEFF Research Database (Denmark)
Palm, Rasmus Berg; Winther, Ole; Laws, Florian
2017-01-01
We present CloudScan; an invoice analysis system that requires zero configuration or upfront annotation. In contrast to previous work, CloudScan does not rely on templates of invoice layout, instead it learns a single global model of invoices that naturally generalizes to unseen invoice layouts....... The model is trained using data automatically extracted from end-user provided feedback. This automatic training data extraction removes the requirement for users to annotate the data precisely. We describe a recurrent neural network model that can capture long range context and compare it to a baseline...... logistic regression model corresponding to the current CloudScan production system. We train and evaluate the system on 8 important fields using a dataset of 326,471 invoices. The recurrent neural network and baseline model achieve 0.891 and 0.887 average F1 scores respectively on seen invoice layouts...
Energy Technology Data Exchange (ETDEWEB)
Han, Seong Ik [Pusan National University, Busan (Korea, Republic of); Jeong, Chan Se; Yang, Soon Yong [University of Ulsan, Ulsan (Korea, Republic of)
2012-04-15
A robust positioning control scheme has been developed using friction parameter observer and recurrent fuzzy neural networks based on the sliding mode control. As a dynamic friction model, the LuGre model is adopted for handling friction compensation because it has been known to capture sufficiently the properties of a nonlinear dynamic friction. A developed friction parameter observer has a simple structure and also well estimates friction parameters of the LuGre friction model. In addition, an approximation method for the system uncertainty is developed using recurrent fuzzy neural networks technology to improve the precision positioning degree. Some simulation and experiment provide the verification on the performance of a proposed robust control scheme.
Directory of Open Access Journals (Sweden)
Chih-Hong Lin
2016-06-01
Full Text Available A permanent magnet (PM synchronous generator system driven by wind turbine (WT, connected with smart grid via AC-DC converter and DC-AC converter, are controlled by the novel recurrent Chebyshev neural network (NN and amended particle swarm optimization (PSO to regulate output power and output voltage in two power converters in this study. Because a PM synchronous generator system driven by WT is an unknown non-linear and time-varying dynamic system, the on-line training novel recurrent Chebyshev NN control system is developed to regulate DC voltage of the AC-DC converter and AC voltage of the DC-AC converter connected with smart grid. Furthermore, the variable learning rate of the novel recurrent Chebyshev NN is regulated according to discrete-type Lyapunov function for improving the control performance and enhancing convergent speed. Finally, some experimental results are shown to verify the effectiveness of the proposed control method for a WT driving a PM synchronous generator system in smart grid.
Online Signature Verification using Recurrent Neural Network and Length-normalized Path Signature
Lai, Songxuan; Jin, Lianwen; Yang, Weixin
2017-01-01
Inspired by the great success of recurrent neural networks (RNNs) in sequential modeling, we introduce a novel RNN system to improve the performance of online signature verification. The training objective is to directly minimize intra-class variations and to push the distances between skilled forgeries and genuine samples above a given threshold. By back-propagating the training signals, our RNN network produced discriminative features with desired metrics. Additionally, we propose a novel d...
Complex Dynamical Network Control for Trajectory Tracking Using Delayed Recurrent Neural Networks
Directory of Open Access Journals (Sweden)
Jose P. Perez
2014-01-01
Full Text Available In this paper, the problem of trajectory tracking is studied. Based on the V-stability and Lyapunov theory, a control law that achieves the global asymptotic stability of the tracking error between a delayed recurrent neural network and a complex dynamical network is obtained. To illustrate the analytic results, we present a tracking simulation of a dynamical network with each node being just one Lorenz’s dynamical system and three identical Chen’s dynamical systems.
Online Sequence Training of Recurrent Neural Networks with Connectionist Temporal Classification
Hwang, Kyuyeon; Sung, Wonyong
2015-01-01
Connectionist temporal classification (CTC) based supervised sequence training of recurrent neural networks (RNNs) has shown great success in many machine learning areas including end-to-end speech and handwritten character recognition. For the CTC training, however, it is required to unroll (or unfold) the RNN by the length of an input sequence. This unrolling requires a lot of memory and hinders a small footprint implementation of online learning or adaptation. Furthermore, the length of tr...
International Nuclear Information System (INIS)
Cadini, F.; Zio, E.; Pedroni, N.
2007-01-01
In this paper, a locally recurrent neural network (LRNN) is employed for approximating the temporal evolution of a nonlinear dynamic system model of a simplified nuclear reactor. To this aim, an infinite impulse response multi-layer perceptron (IIR-MLP) is trained according to a recursive back-propagation (RBP) algorithm. The network nodes contain internal feedback paths and their connections are realized by means of IIR synaptic filters, which provide the LRNN with the necessary system state memory
Some new results for recurrent neural networks with varying-time coefficients and delays
International Nuclear Information System (INIS)
Jiang Haijun; Teng Zhidong
2005-01-01
In this Letter, we consider the recurrent neural networks with varying-time coefficients and delays. By constructing new Lyapunov functional, introducing ingeniously many real parameters and applying the technique of Young inequality, we establish a series of criteria on the boundedness, global exponential stability and the existence of periodic solutions. In these criteria, we do not require that the response functions are differentiable, bounded and monotone nondecreasing. Some previous works are improved and extended
Li, Xiangang; Wu, Xihong
2014-01-01
Long short-term memory (LSTM) based acoustic modeling methods have recently been shown to give state-of-the-art performance on some speech recognition tasks. To achieve a further performance improvement, in this research, deep extensions on LSTM are investigated considering that deep hierarchical model has turned out to be more efficient than a shallow one. Motivated by previous research on constructing deep recurrent neural networks (RNNs), alternative deep LSTM architectures are proposed an...
Yin, Zi; Chang, Keng-hao; Zhang, Ruofei
2017-01-01
Information extraction and user intention identification are central topics in modern query understanding and recommendation systems. In this paper, we propose DeepProbe, a generic information-directed interaction framework which is built around an attention-based sequence to sequence (seq2seq) recurrent neural network. DeepProbe can rephrase, evaluate, and even actively ask questions, leveraging the generative ability and likelihood estimation made possible by seq2seq models. DeepProbe makes...
A Heuristic Approach to Intra-Brain Communications Using Chaos in a Recurrent Neural Network Model
Soma, Ken-ichiro; Mori, Ryota; Sato, Ryuichi; Nara, Shigetoshi
2011-09-01
To approach functional roles of chaos in brain, a heuristic model to consider mechanisms of intra-brain communications is proposed. The key idea is to use chaos in firing pattern dynamics of a recurrent neural network consisting of birary state neurons, as propagation medium of pulse signals. Computer experiments and numerical methods are introduced to evaluate signal transport characteristics by calculating correlation functions between sending neurons and receiving neurons of pulse signals.
Encoding Time in Feedforward Trajectories of a Recurrent Neural Network Model.
Hardy, N F; Buonomano, Dean V
2018-02-01
Brain activity evolves through time, creating trajectories of activity that underlie sensorimotor processing, behavior, and learning and memory. Therefore, understanding the temporal nature of neural dynamics is essential to understanding brain function and behavior. In vivo studies have demonstrated that sequential transient activation of neurons can encode time. However, it remains unclear whether these patterns emerge from feedforward network architectures or from recurrent networks and, furthermore, what role network structure plays in timing. We address these issues using a recurrent neural network (RNN) model with distinct populations of excitatory and inhibitory units. Consistent with experimental data, a single RNN could autonomously produce multiple functionally feedforward trajectories, thus potentially encoding multiple timed motor patterns lasting up to several seconds. Importantly, the model accounted for Weber's law, a hallmark of timing behavior. Analysis of network connectivity revealed that efficiency-a measure of network interconnectedness-decreased as the number of stored trajectories increased. Additionally, the balance of excitation (E) and inhibition (I) shifted toward excitation during each unit's activation time, generating the prediction that observed sequential activity relies on dynamic control of the E/I balance. Our results establish for the first time that the same RNN can generate multiple functionally feedforward patterns of activity as a result of dynamic shifts in the E/I balance imposed by the connectome of the RNN. We conclude that recurrent network architectures account for sequential neural activity, as well as for a fundamental signature of timing behavior: Weber's law.
International Nuclear Information System (INIS)
Hajihosseini, Amirhossein; Maleki, Farzaneh; Rokni Lamooki, Gholam Reza
2011-01-01
Highlights: → We construct a recurrent neural network by generalizing a specific n-neuron network. → Several codimension 1 and 2 bifurcations take place in the newly constructed network. → The newly constructed network has higher capabilities to learn periodic signals. → The normal form theorem is applied to investigate dynamics of the network. → A series of bifurcation diagrams is given to support theoretical results. - Abstract: A class of recurrent neural networks is constructed by generalizing a specific class of n-neuron networks. It is shown that the newly constructed network experiences generic pitchfork and Hopf codimension one bifurcations. It is also proved that the emergence of generic Bogdanov-Takens, pitchfork-Hopf and Hopf-Hopf codimension two, and the degenerate Bogdanov-Takens bifurcation points in the parameter space is possible due to the intersections of codimension one bifurcation curves. The occurrence of bifurcations of higher codimensions significantly increases the capability of the newly constructed recurrent neural network to learn broader families of periodic signals.
Kumar, Rajesh; Srivastava, Smriti; Gupta, J R P
2017-03-01
In this paper adaptive control of nonlinear dynamical systems using diagonal recurrent neural network (DRNN) is proposed. The structure of DRNN is a modification of fully connected recurrent neural network (FCRNN). Presence of self-recurrent neurons in the hidden layer of DRNN gives it an ability to capture the dynamic behaviour of the nonlinear plant under consideration (to be controlled). To ensure stability, update rules are developed using lyapunov stability criterion. These rules are then used for adjusting the various parameters of DRNN. The responses of plants obtained with DRNN are compared with those obtained when multi-layer feed forward neural network (MLFFNN) is used as a controller. Also, in example 4, FCRNN is also investigated and compared with DRNN and MLFFNN. Robustness of the proposed control scheme is also tested against parameter variations and disturbance signals. Four simulation examples including one-link robotic manipulator and inverted pendulum are considered on which the proposed controller is applied. The results so obtained show the superiority of DRNN over MLFFNN as a controller. Copyright © 2017 ISA. Published by Elsevier Ltd. All rights reserved.
Medical Concept Normalization in Social Media Posts with Recurrent Neural Networks.
Tutubalina, Elena; Miftahutdinov, Zulfat; Nikolenko, Sergey; Malykh, Valentin
2018-06-12
Text mining of scientific libraries and social media has already proven itself as a reliable tool for drug repurposing and hypothesis generation. The task of mapping a disease mention to a concept in a controlled vocabulary, typically to the standard thesaurus in the Unified Medical Language System (UMLS), is known as medical concept normalization. This task is challenging due to the differences in the use of medical terminology between health care professionals and social media texts coming from the lay public. To bridge this gap, we use sequence learning with recurrent neural networks and semantic representation of one- or multi-word expressions: we develop end-to-end architectures directly tailored to the task, including bidirectional Long Short-Term Memory, Gated Recurrent Units with an attention mechanism, and additional semantic similarity features based on UMLS. Our evaluation against a standard benchmark shows that recurrent neural networks improve results over an effective baseline for classification based on convolutional neural networks. A qualitative examination of mentions discovered in a dataset of user reviews collected from popular online health information platforms as well as a quantitative evaluation both show improvements in the semantic representation of health-related expressions in social media. Copyright © 2018. Published by Elsevier Inc.
Babaei, Sepideh; Geranmayeh, Amir; Seyyedsalehi, Seyyed Ali
2010-12-01
The supervised learning of recurrent neural networks well-suited for prediction of protein secondary structures from the underlying amino acids sequence is studied. Modular reciprocal recurrent neural networks (MRR-NN) are proposed to model the strong correlations between adjacent secondary structure elements. Besides, a multilayer bidirectional recurrent neural network (MBR-NN) is introduced to capture the long-range intramolecular interactions between amino acids in formation of the secondary structure. The final modular prediction system is devised based on the interactive integration of the MRR-NN and the MBR-NN structures to arbitrarily engage the neighboring effects of the secondary structure types concurrent with memorizing the sequential dependencies of amino acids along the protein chain. The advanced combined network augments the percentage accuracy (Q₃) to 79.36% and boosts the segment overlap (SOV) up to 70.09% when tested on the PSIPRED dataset in three-fold cross-validation. Copyright © 2010 Elsevier Ireland Ltd. All rights reserved.
Directory of Open Access Journals (Sweden)
Hyoung‐Gook Kim
2017-12-01
Full Text Available Recently, deep recurrent neural networks have achieved great success in various machine learning tasks, and have also been applied for sound event detection. The detection of temporally overlapping sound events in realistic environments is much more challenging than in monophonic detection problems. In this paper, we present an approach to improve the accuracy of polyphonic sound event detection in multichannel audio based on gated recurrent neural networks in combination with auditory spectral features. In the proposed method, human hearing perception‐based spatial and spectral‐domain noise‐reduced harmonic features are extracted from multichannel audio and used as high‐resolution spectral inputs to train gated recurrent neural networks. This provides a fast and stable convergence rate compared to long short‐term memory recurrent neural networks. Our evaluation reveals that the proposed method outperforms the conventional approaches.
Neural correlates of working memory in first episode and recurrent depression: An fMRI study.
Yüksel, Dilara; Dietsche, Bruno; Konrad, Carsten; Dannlowski, Udo; Kircher, Tilo; Krug, Axel
2018-06-08
Patients suffering from major depressive disorder (MDD) show deficits in working memory (WM) performance accompanied by bilateral fronto-parietal BOLD signal changes. It is unclear whether patients with a first depressive episode (FDE) exhibit the same signal changes as patients with recurrent depressive episodes (RDE). We investigated seventy-four MDD inpatients (48 RDE, 26 FDE) and 74 healthy control (HC) subjects performing an n-back WM task (0-back, 2-back, 3-back condition) in a 3T-fMRI. FMRI analyses revealed deviating BOLD signal in MDD in the thalamus (0-back vs. 2-back), the angular gyrus (0-back vs. 3-back), and the superior frontal gyrus (2-back vs. 3-back). Further effects were observed between RDE vs. FDE. Thus, RDE displayed differing neural activation in the middle frontal gyrus (2-back vs. 3-back), the inferior frontal gyrus, and the precentral gyrus (0-back vs. 2-back). In addition, both HC and FDE indicated a linear activation trend depending on task complexity. Although we failed to find behavioral differences between the groups, results suggest differing BOLD signal in fronto-parietal brain regions in MDD vs. HC, and in RDE vs. FDE. Moreover, both HC and FDE show similar trends in activation shapes. This indicates a link between levels of complexity-dependent activation in fronto-parietal brain regions and the stage of MDD. We therefore assume that load-dependent BOLD signal during WM is impaired in MDD, and that it is particularly affected in RDE. We also suspect neurobiological compensatory mechanisms of the reported brain regions in (working) memory functioning. Copyright © 2018 Elsevier Inc. All rights reserved.
Budzinski, R. C.; Boaretto, B. R. R.; Prado, T. L.; Lopes, S. R.
2017-07-01
We study the stability of asymptotic states displayed by a complex neural network. We focus on the loss of stability of a stationary state of networks using recurrence quantifiers as tools to diagnose local and global stabilities as well as the multistability of a coupled neural network. Numerical simulations of a neural network composed of 1024 neurons in a small-world connection scheme are performed using the model of Braun et al. [Int. J. Bifurcation Chaos 08, 881 (1998), 10.1142/S0218127498000681], which is a modified model from the Hodgkin-Huxley model [J. Phys. 117, 500 (1952)]. To validate the analyses, the results are compared with those produced by Kuramoto's order parameter [Chemical Oscillations, Waves, and Turbulence (Springer-Verlag, Berlin Heidelberg, 1984)]. We show that recurrence tools making use of just integrated signals provided by the networks, such as local field potential (LFP) (LFP signals) or mean field values bring new results on the understanding of neural behavior occurring before the synchronization states. In particular we show the occurrence of different stationary and nonstationarity asymptotic states.
Interpretation of correlated neural variability from models of feed-forward and recurrent circuits
2018-01-01
Neural populations respond to the repeated presentations of a sensory stimulus with correlated variability. These correlations have been studied in detail, with respect to their mechanistic origin, as well as their influence on stimulus discrimination and on the performance of population codes. A number of theoretical studies have endeavored to link network architecture to the nature of the correlations in neural activity. Here, we contribute to this effort: in models of circuits of stochastic neurons, we elucidate the implications of various network architectures—recurrent connections, shared feed-forward projections, and shared gain fluctuations—on the stimulus dependence in correlations. Specifically, we derive mathematical relations that specify the dependence of population-averaged covariances on firing rates, for different network architectures. In turn, these relations can be used to analyze data on population activity. We examine recordings from neural populations in mouse auditory cortex. We find that a recurrent network model with random effective connections captures the observed statistics. Furthermore, using our circuit model, we investigate the relation between network parameters, correlations, and how well different stimuli can be discriminated from one another based on the population activity. As such, our approach allows us to relate properties of the neural circuit to information processing. PMID:29408930
Interpretation of correlated neural variability from models of feed-forward and recurrent circuits.
Directory of Open Access Journals (Sweden)
Volker Pernice
2018-02-01
Full Text Available Neural populations respond to the repeated presentations of a sensory stimulus with correlated variability. These correlations have been studied in detail, with respect to their mechanistic origin, as well as their influence on stimulus discrimination and on the performance of population codes. A number of theoretical studies have endeavored to link network architecture to the nature of the correlations in neural activity. Here, we contribute to this effort: in models of circuits of stochastic neurons, we elucidate the implications of various network architectures-recurrent connections, shared feed-forward projections, and shared gain fluctuations-on the stimulus dependence in correlations. Specifically, we derive mathematical relations that specify the dependence of population-averaged covariances on firing rates, for different network architectures. In turn, these relations can be used to analyze data on population activity. We examine recordings from neural populations in mouse auditory cortex. We find that a recurrent network model with random effective connections captures the observed statistics. Furthermore, using our circuit model, we investigate the relation between network parameters, correlations, and how well different stimuli can be discriminated from one another based on the population activity. As such, our approach allows us to relate properties of the neural circuit to information processing.
International Nuclear Information System (INIS)
Cao Jiacong; Lin Xingchun
2008-01-01
An accurate forecast of solar irradiation is required for various solar energy applications and environmental impact analyses in recent years. Comparatively, various irradiation forecast models based on artificial neural networks (ANN) perform much better in accuracy than many conventional prediction models. However, the forecast precision of most existing ANN based forecast models has not been satisfactory to researchers and engineers so far, and the generalization capability of these networks needs further improving. Combining the prominent dynamic properties of a recurrent neural network (RNN) with the enhanced ability of a wavelet neural network (WNN) in mapping nonlinear functions, a diagonal recurrent wavelet neural network (DRWNN) is newly established in this paper to perform fine forecasting of hourly and daily global solar irradiance. Some additional steps, e.g. applying historical information of cloud cover to sample data sets and the cloud cover from the weather forecast to network input, are adopted to help enhance the forecast precision. Besides, a specially scheduled two phase training algorithm is adopted. As examples, both hourly and daily irradiance forecasts are completed using sample data sets in Shanghai and Macau, and comparisons between irradiation models show that the DRWNN models are definitely more accurate
Reward-based training of recurrent neural networks for cognitive and value-based tasks.
Song, H Francis; Yang, Guangyu R; Wang, Xiao-Jing
2017-01-13
Trained neural network models, which exhibit features of neural activity recorded from behaving animals, may provide insights into the circuit mechanisms of cognitive functions through systematic analysis of network activity and connectivity. However, in contrast to the graded error signals commonly used to train networks through supervised learning, animals learn from reward feedback on definite actions through reinforcement learning. Reward maximization is particularly relevant when optimal behavior depends on an animal's internal judgment of confidence or subjective preferences. Here, we implement reward-based training of recurrent neural networks in which a value network guides learning by using the activity of the decision network to predict future reward. We show that such models capture behavioral and electrophysiological findings from well-known experimental paradigms. Our work provides a unified framework for investigating diverse cognitive and value-based computations, and predicts a role for value representation that is essential for learning, but not executing, a task.
Robust recurrent neural network modeling for software fault detection and correction prediction
International Nuclear Information System (INIS)
Hu, Q.P.; Xie, M.; Ng, S.H.; Levitin, G.
2007-01-01
Software fault detection and correction processes are related although different, and they should be studied together. A practical approach is to apply software reliability growth models to model fault detection, and fault correction process is assumed to be a delayed process. On the other hand, the artificial neural networks model, as a data-driven approach, tries to model these two processes together with no assumptions. Specifically, feedforward backpropagation networks have shown their advantages over analytical models in fault number predictions. In this paper, the following approach is explored. First, recurrent neural networks are applied to model these two processes together. Within this framework, a systematic networks configuration approach is developed with genetic algorithm according to the prediction performance. In order to provide robust predictions, an extra factor characterizing the dispersion of prediction repetitions is incorporated into the performance function. Comparisons with feedforward neural networks and analytical models are developed with respect to a real data set
Neural network-based robust actuator fault diagnosis for a non-linear multi-tank system.
Mrugalski, Marcin; Luzar, Marcel; Pazera, Marcin; Witczak, Marcin; Aubrun, Christophe
2016-03-01
The paper is devoted to the problem of the robust actuator fault diagnosis of the dynamic non-linear systems. In the proposed method, it is assumed that the diagnosed system can be modelled by the recurrent neural network, which can be transformed into the linear parameter varying form. Such a system description allows developing the designing scheme of the robust unknown input observer within H∞ framework for a class of non-linear systems. The proposed approach is designed in such a way that a prescribed disturbance attenuation level is achieved with respect to the actuator fault estimation error, while guaranteeing the convergence of the observer. The application of the robust unknown input observer enables actuator fault estimation, which allows applying the developed approach to the fault tolerant control tasks. Copyright © 2016 ISA. Published by Elsevier Ltd. All rights reserved.
DEFF Research Database (Denmark)
Dasgupta, Sakyasingha; Goldschmidt, Dennis; Wörgötter, Florentin
2015-01-01
here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures) with the underlying neural mechanisms. The neural mechanisms consist of (1) central pattern generator based control for generating basic rhythmic patterns and coordinated......Walking animals, like stick insects, cockroaches or ants, demonstrate a fascinating range of locomotive abilities and complex behaviors. The locomotive behaviors can consist of a variety of walking patterns along with adaptation that allow the animals to deal with changes in environmental...... conditions, like uneven terrains, gaps, obstacles etc. Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions. While the biomechanics helps maintain flexibility and sustain...
Lee, S R; Choi, S B
2017-06-01
Pediatric laparoscopic herniorrhaphy has rare complications, but recurrence might occur. The purpose of this manuscript is to evaluate the efficacy of linear suture technique of laparoscopic pediatric herniorrhaphy in reducing recurrences. Laparoscopic surgery was performed on 2223 pediatric patients (under 10 years old) from September 2012 to December 2014 in Damsoyu Hospital, Seoul, Republic of Korea. The causes of recurrence were investigated case by case. The patients were categorized into two groups according to the suture method used in closing the hernia orifice: Group 1 (purse-string suture, 1009 patients) and Group 2 (linear suture, 1214 patients). There were 1413 (63.6%) male and 810 (36.4%) female patients. Mean age was 30.5 ± 29.2 months. A significantly higher proportion of male patients, contralateral patent processus vaginalis, and less proportion of recurrence were observed in Group 2. There were ten cases of recurrence in Group 1 because the internal ring suture could not endure the tension. One recurrence occurred in Group 2. The suture technique and age were found to be a significant risk factor for recurrence. Linear suture technique had a lower recurrence rate (odds ratio = 0.07, with 95% confidence interval 0.01-0.53, and p = 0.004). Purse-string suture technique causes significantly higher occurrence of hernia recurrences than linear suture technique. Linear suture technique can reduce recurrence by increasing the endurance to tension around the internal ring by distributing pressure to a wider area along the linear suture line. Linear suture technique can effectively reduce recurrence in pediatric inguinal herniorrhaphy.
Brain Dynamics in Predicting Driving Fatigue Using a Recurrent Self-Evolving Fuzzy Neural Network.
Liu, Yu-Ting; Lin, Yang-Yin; Wu, Shang-Lin; Chuang, Chun-Hsiang; Lin, Chin-Teng
2016-02-01
This paper proposes a generalized prediction system called a recurrent self-evolving fuzzy neural network (RSEFNN) that employs an on-line gradient descent learning rule to address the electroencephalography (EEG) regression problem in brain dynamics for driving fatigue. The cognitive states of drivers significantly affect driving safety; in particular, fatigue driving, or drowsy driving, endangers both the individual and the public. For this reason, the development of brain-computer interfaces (BCIs) that can identify drowsy driving states is a crucial and urgent topic of study. Many EEG-based BCIs have been developed as artificial auxiliary systems for use in various practical applications because of the benefits of measuring EEG signals. In the literature, the efficacy of EEG-based BCIs in recognition tasks has been limited by low resolutions. The system proposed in this paper represents the first attempt to use the recurrent fuzzy neural network (RFNN) architecture to increase adaptability in realistic EEG applications to overcome this bottleneck. This paper further analyzes brain dynamics in a simulated car driving task in a virtual-reality environment. The proposed RSEFNN model is evaluated using the generalized cross-subject approach, and the results indicate that the RSEFNN is superior to competing models regardless of the use of recurrent or nonrecurrent structures.
Deep Recurrent Neural Network-Based Autoencoders for Acoustic Novelty Detection
Directory of Open Access Journals (Sweden)
Erik Marchi
2017-01-01
Full Text Available In the emerging field of acoustic novelty detection, most research efforts are devoted to probabilistic approaches such as mixture models or state-space models. Only recent studies introduced (pseudo-generative models for acoustic novelty detection with recurrent neural networks in the form of an autoencoder. In these approaches, auditory spectral features of the next short term frame are predicted from the previous frames by means of Long-Short Term Memory recurrent denoising autoencoders. The reconstruction error between the input and the output of the autoencoder is used as activation signal to detect novel events. There is no evidence of studies focused on comparing previous efforts to automatically recognize novel events from audio signals and giving a broad and in depth evaluation of recurrent neural network-based autoencoders. The present contribution aims to consistently evaluate our recent novel approaches to fill this white spot in the literature and provide insight by extensive evaluations carried out on three databases: A3Novelty, PASCAL CHiME, and PROMETHEUS. Besides providing an extensive analysis of novel and state-of-the-art methods, the article shows how RNN-based autoencoders outperform statistical approaches up to an absolute improvement of 16.4% average F-measure over the three databases.
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Ordóñez, Francisco Javier; Roggen, Daniel
2016-01-01
Human activity recognition (HAR) tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we propose a generic deep framework for activity recognition based on convolutional and LSTM recurrent units, which: (i) is suitable for multimodal wearable sensors; (ii) can perform sensor fusion naturally; (iii) does not require expert knowledge in designing features; and (iv) explicitly models the temporal dynamics of feature activations. We evaluate our framework on two datasets, one of which has been used in a public activity recognition challenge. Our results show that our framework outperforms competing deep non-recurrent networks on the challenge dataset by 4% on average; outperforming some of the previous reported results by up to 9%. Our results show that the framework can be applied to homogeneous sensor modalities, but can also fuse multimodal sensors to improve performance. We characterise key architectural hyperparameters’ influence on performance to provide insights about their optimisation. PMID:26797612
Deep Recurrent Neural Network-Based Autoencoders for Acoustic Novelty Detection.
Marchi, Erik; Vesperini, Fabio; Squartini, Stefano; Schuller, Björn
2017-01-01
In the emerging field of acoustic novelty detection, most research efforts are devoted to probabilistic approaches such as mixture models or state-space models. Only recent studies introduced (pseudo-)generative models for acoustic novelty detection with recurrent neural networks in the form of an autoencoder. In these approaches, auditory spectral features of the next short term frame are predicted from the previous frames by means of Long-Short Term Memory recurrent denoising autoencoders. The reconstruction error between the input and the output of the autoencoder is used as activation signal to detect novel events. There is no evidence of studies focused on comparing previous efforts to automatically recognize novel events from audio signals and giving a broad and in depth evaluation of recurrent neural network-based autoencoders. The present contribution aims to consistently evaluate our recent novel approaches to fill this white spot in the literature and provide insight by extensive evaluations carried out on three databases: A3Novelty, PASCAL CHiME, and PROMETHEUS. Besides providing an extensive analysis of novel and state-of-the-art methods, the article shows how RNN-based autoencoders outperform statistical approaches up to an absolute improvement of 16.4% average F -measure over the three databases.
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Directory of Open Access Journals (Sweden)
Francisco Javier Ordóñez
2016-01-01
Full Text Available Human activity recognition (HAR tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we propose a generic deep framework for activity recognition based on convolutional and LSTM recurrent units, which: (i is suitable for multimodal wearable sensors; (ii can perform sensor fusion naturally; (iii does not require expert knowledge in designing features; and (iv explicitly models the temporal dynamics of feature activations. We evaluate our framework on two datasets, one of which has been used in a public activity recognition challenge. Our results show that our framework outperforms competing deep non-recurrent networks on the challenge dataset by 4% on average; outperforming some of the previous reported results by up to 9%. Our results show that the framework can be applied to homogeneous sensor modalities, but can also fuse multimodal sensors to improve performance. We characterise key architectural hyperparameters’ influence on performance to provide insights about their optimisation.
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition.
Ordóñez, Francisco Javier; Roggen, Daniel
2016-01-18
Human activity recognition (HAR) tasks have traditionally been solved using engineered features obtained by heuristic processes. Current research suggests that deep convolutional neural networks are suited to automate feature extraction from raw sensor inputs. However, human activities are made of complex sequences of motor movements, and capturing this temporal dynamics is fundamental for successful HAR. Based on the recent success of recurrent neural networks for time series domains, we propose a generic deep framework for activity recognition based on convolutional and LSTM recurrent units, which: (i) is suitable for multimodal wearable sensors; (ii) can perform sensor fusion naturally; (iii) does not require expert knowledge in designing features; and (iv) explicitly models the temporal dynamics of feature activations. We evaluate our framework on two datasets, one of which has been used in a public activity recognition challenge. Our results show that our framework outperforms competing deep non-recurrent networks on the challenge dataset by 4% on average; outperforming some of the previous reported results by up to 9%. Our results show that the framework can be applied to homogeneous sensor modalities, but can also fuse multimodal sensors to improve performance. We characterise key architectural hyperparameters' influence on performance to provide insights about their optimisation.
Cao, Renzhi; Freitas, Colton; Chan, Leong; Sun, Miao; Jiang, Haiqing; Chen, Zhangxin
2017-10-17
With the development of next generation sequencing techniques, it is fast and cheap to determine protein sequences but relatively slow and expensive to extract useful information from protein sequences because of limitations of traditional biological experimental techniques. Protein function prediction has been a long standing challenge to fill the gap between the huge amount of protein sequences and the known function. In this paper, we propose a novel method to convert the protein function problem into a language translation problem by the new proposed protein sequence language "ProLan" to the protein function language "GOLan", and build a neural machine translation model based on recurrent neural networks to translate "ProLan" language to "GOLan" language. We blindly tested our method by attending the latest third Critical Assessment of Function Annotation (CAFA 3) in 2016, and also evaluate the performance of our methods on selected proteins whose function was released after CAFA competition. The good performance on the training and testing datasets demonstrates that our new proposed method is a promising direction for protein function prediction. In summary, we first time propose a method which converts the protein function prediction problem to a language translation problem and applies a neural machine translation model for protein function prediction.
Directory of Open Access Journals (Sweden)
Eduard eGrinke
2015-10-01
Full Text Available Walking animals, like insects, with little neural computing can effectively perform complex behaviors. They can walk around their environment, escape from corners/deadlocks, and avoid or climb over obstacles. While performing all these behaviors, they can also adapt their movements to deal with an unknown situation. As a consequence, they successfully navigate through their complex environment. The versatile and adaptive abilities are the result of an integration of several ingredients embedded in their sensorimotor loop. Biological studies reveal that the ingredients include neural dynamics, plasticity, sensory feedback, and biomechanics. Generating such versatile and adaptive behaviors for a walking robot is a challenging task. In this study, we present a bio-inspired approach to solve this task. Specifically, the approach combines neural mechanisms with plasticity, sensory feedback, and biomechanics. The neural mechanisms consist of adaptive neural sensory processing and modular neural locomotion control. The sensory processing is based on a small recurrent network consisting of two fully connected neurons. Online correlation-based learning with synaptic scaling is applied to adequately change the connections of the network. By doing so, we can effectively exploit neural dynamics (i.e., hysteresis effects and single attractors in the network to generate different turning angles with short-term memory for a biomechanical walking robot. The turning information is transmitted as descending steering signals to the locomotion control which translates the signals into motor actions. As a result, the robot can walk around and adapt its turning angle for avoiding obstacles in different situations as well as escaping from sharp corners or deadlocks. Using backbone joint control embedded in the locomotion control allows the robot to climb over small obstacles. Consequently, it can successfully explore and navigate in complex environments.
International Nuclear Information System (INIS)
Wang, Jie; Wang, Jun
2016-01-01
In an attempt to improve the forecasting accuracy of crude oil price fluctuations, a new neural network architecture is established in this work which combines Multilayer perception and ERNN (Elman recurrent neural networks) with stochastic time effective function. ERNN is a time-varying predictive control system and is developed with the ability to keep memory of recent events in order to predict future output. The stochastic time effective function represents that the recent information has a stronger effect for the investors than the old information. With the established model the empirical research has a good performance in testing the predictive effects on four different time series indices. Compared to other models, the present model is possible to evaluate data from 1990s to today with extreme accuracy and speedy. The applied CID (complexity invariant distance) analysis and multiscale CID analysis, are provided as the new useful measures to evaluate a better predicting ability of the proposed model than other traditional models. - Highlights: • A new forecasting model is developed by a random Elman recurrent neural network. • The forecasting accuracy of crude oil price fluctuations is improved by the model. • The forecasting results of the proposed model are more accurate than compared models. • Two new distance analysis methods are applied to confirm the predicting results.
Hou, Runmin; Wang, Li; Gao, Qiang; Hou, Yuanglong; Wang, Chao
2017-09-01
This paper proposes a novel indirect adaptive fuzzy wavelet neural network (IAFWNN) to control the nonlinearity, wide variations in loads, time-variation and uncertain disturbance of the ac servo system. In the proposed approach, the self-recurrent wavelet neural network (SRWNN) is employed to construct an adaptive self-recurrent consequent part for each fuzzy rule of TSK fuzzy model. For the IAFWNN controller, the online learning algorithm is based on back propagation (BP) algorithm. Moreover, an improved particle swarm optimization (IPSO) is used to adapt the learning rate. The aid of an adaptive SRWNN identifier offers the real-time gradient information to the adaptive fuzzy wavelet neural controller to overcome the impact of parameter variations, load disturbances and other uncertainties effectively, and has a good dynamic. The asymptotical stability of the system is guaranteed by using the Lyapunov method. The result of the simulation and the prototype test prove that the proposed are effective and suitable. Copyright © 2017. Published by Elsevier Ltd.
International Nuclear Information System (INIS)
Song, Qiankun; Wang, Zidong
2007-01-01
In this Letter, the analysis problem for the existence and stability of periodic solutions is investigated for a class of general discrete-time recurrent neural networks with time-varying delays. For the neural networks under study, a generalized activation function is considered, and the traditional assumptions on the boundedness, monotony and differentiability of the activation functions are removed. By employing the latest free-weighting matrix method, an appropriate Lyapunov-Krasovskii functional is constructed and several sufficient conditions are established to ensure the existence, uniqueness, and globally exponential stability of the periodic solution for the addressed neural network. The conditions are dependent on both the lower bound and upper bound of the time-varying time delays. Furthermore, the conditions are expressed in terms of the linear matrix inequalities (LMIs), which can be checked numerically using the effective LMI toolbox in MATLAB. Two simulation examples are given to show the effectiveness and less conservatism of the proposed criteria
Intelligent fault diagnosis of rolling bearings using an improved deep recurrent neural network
Jiang, Hongkai; Li, Xingqiu; Shao, Haidong; Zhao, Ke
2018-06-01
Traditional intelligent fault diagnosis methods for rolling bearings heavily depend on manual feature extraction and feature selection. For this purpose, an intelligent deep learning method, named the improved deep recurrent neural network (DRNN), is proposed in this paper. Firstly, frequency spectrum sequences are used as inputs to reduce the input size and ensure good robustness. Secondly, DRNN is constructed by the stacks of the recurrent hidden layer to automatically extract the features from the input spectrum sequences. Thirdly, an adaptive learning rate is adopted to improve the training performance of the constructed DRNN. The proposed method is verified with experimental rolling bearing data, and the results confirm that the proposed method is more effective than traditional intelligent fault diagnosis methods.
Terrill, Philip I; Wilson, Stephen J; Suresh, Sadasivam; Cooper, David M; Dakin, Carolyn
2013-05-01
Breathing dynamics vary between infant sleep states, and are likely to exhibit non-linear behaviour. This study applied the non-linear analytical tool recurrence quantification analysis (RQA) to 400 breath interval periods of REM and N-REM sleep, and then using an overlapping moving window. The RQA variables were different between sleep states, with REM radius 150% greater than N-REM radius, and REM laminarity 79% greater than N-REM laminarity. RQA allowed the observation of temporal variations in non-linear breathing dynamics across a night's sleep at 30s resolution, and provides a basis for quantifying changes in complex breathing dynamics with physiology and pathology. Copyright © 2013 Elsevier Ltd. All rights reserved.
Identification of Jets Containing b-Hadrons with Recurrent Neural Networks at the ATLAS Experiment
CERN. Geneva
2017-01-01
A novel b-jet identification algorithm is constructed with a Recurrent Neural Network (RNN) at the ATLAS Experiment. This talk presents the expected performance of the RNN based b-tagging in simulated $t \\bar t$ events. The RNN based b-tagging processes properties of tracks associated to jets which are represented in sequences. In contrast to traditional impact-parameter-based b-tagging algorithms which assume the tracks of jets are independent from each other, RNN based b-tagging can exploit the spatial and kinematic correlations of tracks which are initiated from the same b-hadrons. The neural network nature of the tagging algorithm also allows the flexibility of extending input features to include more track properties than can be effectively used in traditional algorithms.
Directory of Open Access Journals (Sweden)
Wang Chao
2016-03-01
Full Text Available Due to the complexities existing in the electric load simulator, this article develops a high-performance nonlinear adaptive controller to improve the torque tracking performance of the electric load simulator, which mainly consists of an adaptive fuzzy self-recurrent wavelet neural network controller with variable structure (VSFSWC and a complementary controller. The VSFSWC is clearly and easily used for real-time systems and greatly improves the convergence rate and control precision. The complementary controller is designed to eliminate the effect of the approximation error between the proposed neural network controller and the ideal feedback controller without chattering phenomena. Moreover, adaptive learning laws are derived to guarantee the system stability in the sense of the Lyapunov theory. Finally, the hardware-in-the-loop simulations are carried out to verify the feasibility and effectiveness of the proposed algorithms in different working styles.
Identification of serial number on bank card using recurrent neural network
Liu, Li; Huang, Linlin; Xue, Jian
2018-04-01
Identification of serial number on bank card has many applications. Due to the different number printing mode, complex background, distortion in shape, etc., it is quite challenging to achieve high identification accuracy. In this paper, we propose a method using Normalization-Cooperated Gradient Feature (NCGF) and Recurrent Neural Network (RNN) based on Long Short-Term Memory (LSTM) for serial number identification. The NCGF maps the gradient direction elements of original image to direction planes such that the RNN with direction planes as input can recognize numbers more accurately. Taking the advantages of NCGF and RNN, we get 90%digit string recognition accuracy.
Automatic construction of a recurrent neural network based classifier for vehicle passage detection
Burnaev, Evgeny; Koptelov, Ivan; Novikov, German; Khanipov, Timur
2017-03-01
Recurrent Neural Networks (RNNs) are extensively used for time-series modeling and prediction. We propose an approach for automatic construction of a binary classifier based on Long Short-Term Memory RNNs (LSTM-RNNs) for detection of a vehicle passage through a checkpoint. As an input to the classifier we use multidimensional signals of various sensors that are installed on the checkpoint. Obtained results demonstrate that the previous approach to handcrafting a classifier, consisting of a set of deterministic rules, can be successfully replaced by an automatic RNN training on an appropriately labelled data.
Hierarchical Recurrent Neural Hashing for Image Retrieval With Hierarchical Convolutional Features.
Lu, Xiaoqiang; Chen, Yaxiong; Li, Xuelong
Hashing has been an important and effective technology in image retrieval due to its computational efficiency and fast search speed. The traditional hashing methods usually learn hash functions to obtain binary codes by exploiting hand-crafted features, which cannot optimally represent the information of the sample. Recently, deep learning methods can achieve better performance, since deep learning architectures can learn more effective image representation features. However, these methods only use semantic features to generate hash codes by shallow projection but ignore texture details. In this paper, we proposed a novel hashing method, namely hierarchical recurrent neural hashing (HRNH), to exploit hierarchical recurrent neural network to generate effective hash codes. There are three contributions of this paper. First, a deep hashing method is proposed to extensively exploit both spatial details and semantic information, in which, we leverage hierarchical convolutional features to construct image pyramid representation. Second, our proposed deep network can exploit directly convolutional feature maps as input to preserve the spatial structure of convolutional feature maps. Finally, we propose a new loss function that considers the quantization error of binarizing the continuous embeddings into the discrete binary codes, and simultaneously maintains the semantic similarity and balanceable property of hash codes. Experimental results on four widely used data sets demonstrate that the proposed HRNH can achieve superior performance over other state-of-the-art hashing methods.Hashing has been an important and effective technology in image retrieval due to its computational efficiency and fast search speed. The traditional hashing methods usually learn hash functions to obtain binary codes by exploiting hand-crafted features, which cannot optimally represent the information of the sample. Recently, deep learning methods can achieve better performance, since deep
Morillot, Olivier; Likforman-Sulem, Laurence; Grosicki, Emmanuèle
2013-04-01
Many preprocessing techniques have been proposed for isolated word recognition. However, recently, recognition systems have dealt with text blocks and their compound text lines. In this paper, we propose a new preprocessing approach to efficiently correct baseline skew and fluctuations. Our approach is based on a sliding window within which the vertical position of the baseline is estimated. Segmentation of text lines into subparts is, thus, avoided. Experiments conducted on a large publicly available database (Rimes), with a BLSTM (bidirectional long short-term memory) recurrent neural network recognition system, show that our baseline correction approach highly improves performance.
A neural network method for solving a system of linear variational inequalities
International Nuclear Information System (INIS)
Lan Hengyou; Cui Yishun
2009-01-01
In this paper, we transmute the solution for a new system of linear variational inequalities to an equilibrium point of neural networks, and by using analytic technique, some sufficient conditions are presented. Further, the estimation of the exponential convergence rates of the neural networks is investigated. The new and useful results obtained in this paper generalize and improve the corresponding results of recent works.
Nonlinear Model Predictive Control Based on a Self-Organizing Recurrent Neural Network.
Han, Hong-Gui; Zhang, Lu; Hou, Ying; Qiao, Jun-Fei
2016-02-01
A nonlinear model predictive control (NMPC) scheme is developed in this paper based on a self-organizing recurrent radial basis function (SR-RBF) neural network, whose structure and parameters are adjusted concurrently in the training process. The proposed SR-RBF neural network is represented in a general nonlinear form for predicting the future dynamic behaviors of nonlinear systems. To improve the modeling accuracy, a spiking-based growing and pruning algorithm and an adaptive learning algorithm are developed to tune the structure and parameters of the SR-RBF neural network, respectively. Meanwhile, for the control problem, an improved gradient method is utilized for the solution of the optimization problem in NMPC. The stability of the resulting control system is proved based on the Lyapunov stability theory. Finally, the proposed SR-RBF neural network-based NMPC (SR-RBF-NMPC) is used to control the dissolved oxygen (DO) concentration in a wastewater treatment process (WWTP). Comparisons with other existing methods demonstrate that the SR-RBF-NMPC can achieve a considerably better model fitting for WWTP and a better control performance for DO concentration.
The super-Turing computational power of plastic recurrent neural networks.
Cabessa, Jérémie; Siegelmann, Hava T
2014-12-01
We study the computational capabilities of a biologically inspired neural model where the synaptic weights, the connectivity pattern, and the number of neurons can evolve over time rather than stay static. Our study focuses on the mere concept of plasticity of the model so that the nature of the updates is assumed to be not constrained. In this context, we show that the so-called plastic recurrent neural networks (RNNs) are capable of the precise super-Turing computational power--as the static analog neural networks--irrespective of whether their synaptic weights are modeled by rational or real numbers, and moreover, irrespective of whether their patterns of plasticity are restricted to bi-valued updates or expressed by any other more general form of updating. Consequently, the incorporation of only bi-valued plastic capabilities in a basic model of RNNs suffices to break the Turing barrier and achieve the super-Turing level of computation. The consideration of more general mechanisms of architectural plasticity or of real synaptic weights does not further increase the capabilities of the networks. These results support the claim that the general mechanism of plasticity is crucially involved in the computational and dynamical capabilities of biological neural networks. They further show that the super-Turing level of computation reflects in a suitable way the capabilities of brain-like models of computation.
Grinke, Eduard; Tetzlaff, Christian; Wörgötter, Florentin; Manoonpong, Poramate
2015-01-01
Walking animals, like insects, with little neural computing can effectively perform complex behaviors. For example, they can walk around their environment, escape from corners/deadlocks, and avoid or climb over obstacles. While performing all these behaviors, they can also adapt their movements to deal with an unknown situation. As a consequence, they successfully navigate through their complex environment. The versatile and adaptive abilities are the result of an integration of several ingredients embedded in their sensorimotor loop. Biological studies reveal that the ingredients include neural dynamics, plasticity, sensory feedback, and biomechanics. Generating such versatile and adaptive behaviors for a many degrees-of-freedom (DOFs) walking robot is a challenging task. Thus, in this study, we present a bio-inspired approach to solve this task. Specifically, the approach combines neural mechanisms with plasticity, exteroceptive sensory feedback, and biomechanics. The neural mechanisms consist of adaptive neural sensory processing and modular neural locomotion control. The sensory processing is based on a small recurrent neural network consisting of two fully connected neurons. Online correlation-based learning with synaptic scaling is applied to adequately change the connections of the network. By doing so, we can effectively exploit neural dynamics (i.e., hysteresis effects and single attractors) in the network to generate different turning angles with short-term memory for a walking robot. The turning information is transmitted as descending steering signals to the neural locomotion control which translates the signals into motor actions. As a result, the robot can walk around and adapt its turning angle for avoiding obstacles in different situations. The adaptation also enables the robot to effectively escape from sharp corners or deadlocks. Using backbone joint control embedded in the the locomotion control allows the robot to climb over small obstacles
International Nuclear Information System (INIS)
Lu Junguo
2008-01-01
In this paper, the global exponential stability and periodicity for a class of reaction-diffusion delayed recurrent neural networks with Dirichlet boundary conditions are addressed by constructing suitable Lyapunov functionals and utilizing some inequality techniques. We first prove global exponential converge to 0 of the difference between any two solutions of the original reaction-diffusion delayed recurrent neural networks with Dirichlet boundary conditions, the existence and uniqueness of equilibrium is the direct results of this procedure. This approach is different from the usually used one where the existence, uniqueness of equilibrium and stability are proved in two separate steps. Furthermore, we prove periodicity of the reaction-diffusion delayed recurrent neural networks with Dirichlet boundary conditions. Sufficient conditions ensuring the global exponential stability and the existence of periodic oscillatory solutions for the reaction-diffusion delayed recurrent neural networks with Dirichlet boundary conditions are given. These conditions are easy to check and have important leading significance in the design and application of reaction-diffusion recurrent neural networks with delays. Finally, two numerical examples are given to show the effectiveness of the obtained results
RM-SORN: a reward-modulated self-organizing recurrent neural network.
Aswolinskiy, Witali; Pipa, Gordon
2015-01-01
Neural plasticity plays an important role in learning and memory. Reward-modulation of plasticity offers an explanation for the ability of the brain to adapt its neural activity to achieve a rewarded goal. Here, we define a neural network model that learns through the interaction of Intrinsic Plasticity (IP) and reward-modulated Spike-Timing-Dependent Plasticity (STDP). IP enables the network to explore possible output sequences and STDP, modulated by reward, reinforces the creation of the rewarded output sequences. The model is tested on tasks for prediction, recall, non-linear computation, pattern recognition, and sequence generation. It achieves performance comparable to networks trained with supervised learning, while using simple, biologically motivated plasticity rules, and rewarding strategies. The results confirm the importance of investigating the interaction of several plasticity rules in the context of reward-modulated learning and whether reward-modulated self-organization can explain the amazing capabilities of the brain.
Deep recurrent neural network reveals a hierarchy of process memory during dynamic natural vision.
Shi, Junxing; Wen, Haiguang; Zhang, Yizhen; Han, Kuan; Liu, Zhongming
2018-05-01
The human visual cortex extracts both spatial and temporal visual features to support perception and guide behavior. Deep convolutional neural networks (CNNs) provide a computational framework to model cortical representation and organization for spatial visual processing, but unable to explain how the brain processes temporal information. To overcome this limitation, we extended a CNN by adding recurrent connections to different layers of the CNN to allow spatial representations to be remembered and accumulated over time. The extended model, or the recurrent neural network (RNN), embodied a hierarchical and distributed model of process memory as an integral part of visual processing. Unlike the CNN, the RNN learned spatiotemporal features from videos to enable action recognition. The RNN better predicted cortical responses to natural movie stimuli than the CNN, at all visual areas, especially those along the dorsal stream. As a fully observable model of visual processing, the RNN also revealed a cortical hierarchy of temporal receptive window, dynamics of process memory, and spatiotemporal representations. These results support the hypothesis of process memory, and demonstrate the potential of using the RNN for in-depth computational understanding of dynamic natural vision. © 2018 Wiley Periodicals, Inc.
Analysis of recurrent neural networks for short-term energy load forecasting
Di Persio, Luca; Honchar, Oleksandr
2017-11-01
Short-term forecasts have recently gained an increasing attention because of the rise of competitive electricity markets. In fact, short-terms forecast of possible future loads turn out to be fundamental to build efficient energy management strategies as well as to avoid energy wastage. Such type of challenges are difficult to tackle both from a theoretical and applied point of view. Latter tasks require sophisticated methods to manage multidimensional time series related to stochastic phenomena which are often highly interconnected. In the present work we first review novel approaches to energy load forecasting based on recurrent neural network, focusing our attention on long/short term memory architectures (LSTMs). Such type of artificial neural networks have been widely applied to problems dealing with sequential data such it happens, e.g., in socio-economics settings, for text recognition purposes, concerning video signals, etc., always showing their effectiveness to model complex temporal data. Moreover, we consider different novel variations of basic LSTMs, such as sequence-to-sequence approach and bidirectional LSTMs, aiming at providing effective models for energy load data. Last but not least, we test all the described algorithms on real energy load data showing not only that deep recurrent networks can be successfully applied to energy load forecasting, but also that this approach can be extended to other problems based on time series prediction.
Zheng, Jing; Lu, Jiren; Peng, Suping; Jiang, Tianqi
2018-02-01
The conventional arrival pick-up algorithms cannot avoid the manual modification of the parameters for the simultaneous identification of multiple events under different signal-to-noise ratios (SNRs). Therefore, in order to automatically obtain the arrivals of multiple events with high precision under different SNRs, in this study an algorithm was proposed which had the ability to pick up the arrival of microseismic or acoustic emission events based on deep recurrent neural networks. The arrival identification was performed using two important steps, which included a training phase and a testing phase. The training process was mathematically modelled by deep recurrent neural networks using Long Short-Term Memory architecture. During the testing phase, the learned weights were utilized to identify the arrivals through the microseismic/acoustic emission data sets. The data sets were obtained by rock physics experiments of the acoustic emission. In order to obtain the data sets under different SNRs, this study added random noise to the raw experiments' data sets. The results showed that the outcome of the proposed method was able to attain an above 80 per cent hit-rate at SNR 0 dB, and an approximately 70 per cent hit-rate at SNR -5 dB, with an absolute error in 10 sampling points. These results indicated that the proposed method had high selection precision and robustness.
Optimal Formation of Multirobot Systems Based on a Recurrent Neural Network.
Wang, Yunpeng; Cheng, Long; Hou, Zeng-Guang; Yu, Junzhi; Tan, Min
2016-02-01
The optimal formation problem of multirobot systems is solved by a recurrent neural network in this paper. The desired formation is described by the shape theory. This theory can generate a set of feasible formations that share the same relative relation among robots. An optimal formation means that finding one formation from the feasible formation set, which has the minimum distance to the initial formation of the multirobot system. Then, the formation problem is transformed into an optimization problem. In addition, the orientation, scale, and admissible range of the formation can also be considered as the constraints in the optimization problem. Furthermore, if all robots are identical, their positions in the system are exchangeable. Then, each robot does not necessarily move to one specific position in the formation. In this case, the optimal formation problem becomes a combinational optimization problem, whose optimal solution is very hard to obtain. Inspired by the penalty method, this combinational optimization problem can be approximately transformed into a convex optimization problem. Due to the involvement of the Euclidean norm in the distance, the objective function of these optimization problems are nonsmooth. To solve these nonsmooth optimization problems efficiently, a recurrent neural network approach is employed, owing to its parallel computation ability. Finally, some simulations and experiments are given to validate the effectiveness and efficiency of the proposed optimal formation approach.
Using deep recurrent neural network for direct beam solar irradiance cloud screening
Chen, Maosi; Davis, John M.; Liu, Chaoshun; Sun, Zhibin; Zempila, Melina Maria; Gao, Wei
2017-09-01
Cloud screening is an essential procedure for in-situ calibration and atmospheric properties retrieval on (UV-)MultiFilter Rotating Shadowband Radiometer [(UV-)MFRSR]. Previous study has explored a cloud screening algorithm for direct-beam (UV-)MFRSR voltage measurements based on the stability assumption on a long time period (typically a half day or a whole day). To design such an algorithm requires in-depth understanding of radiative transfer and delicate data manipulation. Recent rapid developments on deep neural network and computation hardware have opened a window for modeling complicated End-to-End systems with a standardized strategy. In this study, a multi-layer dynamic bidirectional recurrent neural network is built for determining the cloudiness on each time point with a 17-year training dataset and tested with another 1-year dataset. The dataset is the daily 3-minute cosine corrected voltages, airmasses, and the corresponding cloud/clear-sky labels at two stations of the USDA UV-B Monitoring and Research Program. The results show that the optimized neural network model (3-layer, 250 hidden units, and 80 epochs of training) has an overall test accuracy of 97.87% (97.56% for the Oklahoma site and 98.16% for the Hawaii site). Generally, the neural network model grasps the key concept of the original model to use data in the entire day rather than short nearby measurements to perform cloud screening. A scrutiny of the logits layer suggests that the neural network model automatically learns a way to calculate a quantity similar to total optical depth and finds an appropriate threshold for cloud screening.
Reinforcement Learning of Linking and Tracing Contours in Recurrent Neural Networks
Brosch, Tobias; Neumann, Heiko; Roelfsema, Pieter R.
2015-01-01
The processing of a visual stimulus can be subdivided into a number of stages. Upon stimulus presentation there is an early phase of feedforward processing where the visual information is propagated from lower to higher visual areas for the extraction of basic and complex stimulus features. This is followed by a later phase where horizontal connections within areas and feedback connections from higher areas back to lower areas come into play. In this later phase, image elements that are behaviorally relevant are grouped by Gestalt grouping rules and are labeled in the cortex with enhanced neuronal activity (object-based attention in psychology). Recent neurophysiological studies revealed that reward-based learning influences these recurrent grouping processes, but it is not well understood how rewards train recurrent circuits for perceptual organization. This paper examines the mechanisms for reward-based learning of new grouping rules. We derive a learning rule that can explain how rewards influence the information flow through feedforward, horizontal and feedback connections. We illustrate the efficiency with two tasks that have been used to study the neuronal correlates of perceptual organization in early visual cortex. The first task is called contour-integration and demands the integration of collinear contour elements into an elongated curve. We show how reward-based learning causes an enhancement of the representation of the to-be-grouped elements at early levels of a recurrent neural network, just as is observed in the visual cortex of monkeys. The second task is curve-tracing where the aim is to determine the endpoint of an elongated curve composed of connected image elements. If trained with the new learning rule, neural networks learn to propagate enhanced activity over the curve, in accordance with neurophysiological data. We close the paper with a number of model predictions that can be tested in future neurophysiological and computational studies
A Three-Threshold Learning Rule Approaches the Maximal Capacity of Recurrent Neural Networks.
Directory of Open Access Journals (Sweden)
Alireza Alemi
2015-08-01
Full Text Available Understanding the theoretical foundations of how memories are encoded and retrieved in neural populations is a central challenge in neuroscience. A popular theoretical scenario for modeling memory function is the attractor neural network scenario, whose prototype is the Hopfield model. The model simplicity and the locality of the synaptic update rules come at the cost of a poor storage capacity, compared with the capacity achieved with perceptron learning algorithms. Here, by transforming the perceptron learning rule, we present an online learning rule for a recurrent neural network that achieves near-maximal storage capacity without an explicit supervisory error signal, relying only upon locally accessible information. The fully-connected network consists of excitatory binary neurons with plastic recurrent connections and non-plastic inhibitory feedback stabilizing the network dynamics; the memory patterns to be memorized are presented online as strong afferent currents, producing a bimodal distribution for the neuron synaptic inputs. Synapses corresponding to active inputs are modified as a function of the value of the local fields with respect to three thresholds. Above the highest threshold, and below the lowest threshold, no plasticity occurs. In between these two thresholds, potentiation/depression occurs when the local field is above/below an intermediate threshold. We simulated and analyzed a network of binary neurons implementing this rule and measured its storage capacity for different sizes of the basins of attraction. The storage capacity obtained through numerical simulations is shown to be close to the value predicted by analytical calculations. We also measured the dependence of capacity on the strength of external inputs. Finally, we quantified the statistics of the resulting synaptic connectivity matrix, and found that both the fraction of zero weight synapses and the degree of symmetry of the weight matrix increase with the
Hanson, Jack; Yang, Yuedong; Paliwal, Kuldip; Zhou, Yaoqi
2017-03-01
Capturing long-range interactions between structural but not sequence neighbors of proteins is a long-standing challenging problem in bioinformatics. Recently, long short-term memory (LSTM) networks have significantly improved the accuracy of speech and image classification problems by remembering useful past information in long sequential events. Here, we have implemented deep bidirectional LSTM recurrent neural networks in the problem of protein intrinsic disorder prediction. The new method, named SPOT-Disorder, has steadily improved over a similar method using a traditional, window-based neural network (SPINE-D) in all datasets tested without separate training on short and long disordered regions. Independent tests on four other datasets including the datasets from critical assessment of structure prediction (CASP) techniques and >10 000 annotated proteins from MobiDB, confirmed SPOT-Disorder as one of the best methods in disorder prediction. Moreover, initial studies indicate that the method is more accurate in predicting functional sites in disordered regions. These results highlight the usefulness combining LSTM with deep bidirectional recurrent neural networks in capturing non-local, long-range interactions for bioinformatics applications. SPOT-disorder is available as a web server and as a standalone program at: http://sparks-lab.org/server/SPOT-disorder/index.php . j.hanson@griffith.edu.au or yuedong.yang@griffith.edu.au or yaoqi.zhou@griffith.edu.au. Supplementary data is available at Bioinformatics online. © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
Pérez, Oswaldo; Merchant, Hugo
2018-04-03
Extensive research has described two key features of interval timing. The bias property is associated with accuracy and implies that time is overestimated for short intervals and underestimated for long intervals. The scalar property is linked to precision and states that the variability of interval estimates increases as a function of interval duration. The neural mechanisms behind these properties are not well understood. Here we implemented a recurrent neural network that mimics a cortical ensemble and includes cells that show paired-pulse facilitation and slow inhibitory synaptic currents. The network produces interval selective responses and reproduces both bias and scalar properties when a Bayesian decoder reads its activity. Notably, the interval-selectivity, timing accuracy, and precision of the network showed complex changes as a function of the decay time constants of the modeled synaptic properties and the level of background activity of the cells. These findings suggest that physiological values of the time constants for paired-pulse facilitation and GABAb, as well as the internal state of the network, determine the bias and scalar properties of interval timing. Significant Statement Timing is a fundamental element of complex behavior, including music and language. Temporal processing in a wide variety of contexts shows two primary features: time estimates exhibit a shift towards the mean (the bias property) and are more variable for longer intervals (the scalar property). We implemented a recurrent neural network that includes long-lasting synaptic currents, which can not only produce interval selective responses but also follow the bias and scalar properties. Interestingly, only physiological values of the time constants for paired-pulse facilitation and GABAb, as well as intermediate background activity within the network can reproduce the two key features of interval timing. Copyright © 2018 the authors.
Xia, Peng; Hu, Jie; Peng, Yinghong
2017-10-25
A novel model based on deep learning is proposed to estimate kinematic information for myoelectric control from multi-channel electromyogram (EMG) signals. The neural information of limb movement is embedded in EMG signals that are influenced by all kinds of factors. In order to overcome the negative effects of variability in signals, the proposed model employs the deep architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The EMG signals are transformed to time-frequency frames as the input to the model. The limb movement is estimated by the model that is trained with the gradient descent and backpropagation procedure. We tested the model for simultaneous and proportional estimation of limb movement in eight healthy subjects and compared it with support vector regression (SVR) and CNNs on the same data set. The experimental studies show that the proposed model has higher estimation accuracy and better robustness with respect to time. The combination of CNNs and RNNs can improve the model performance compared with using CNNs alone. The model of deep architecture is promising in EMG decoding and optimization of network structures can increase the accuracy and robustness. © 2017 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.
Jin, Long; Liao, Bolin; Liu, Mei; Xiao, Lin; Guo, Dongsheng; Yan, Xiaogang
2017-01-01
By incorporating the physical constraints in joint space, a different-level simultaneous minimization scheme, which takes both the robot kinematics and robot dynamics into account, is presented and investigated for fault-tolerant motion planning of redundant manipulator in this paper. The scheme is reformulated as a quadratic program (QP) with equality and bound constraints, which is then solved by a discrete-time recurrent neural network. Simulative verifications based on a six-link planar redundant robot manipulator substantiate the efficacy and accuracy of the presented acceleration fault-tolerant scheme, the resultant QP and the corresponding discrete-time recurrent neural network.
Naikwad, S. N.; Dudul, S. V.
2009-01-01
A focused time lagged recurrent neural network (FTLR NN) with gamma memory filter is designed to learn the subtle complex dynamics of a typical CSTR process. Continuous stirred tank reactor exhibits complex nonlinear operations where reaction is exothermic. It is noticed from literature review that process control of CSTR using neuro-fuzzy systems was attempted by many, but optimal neural network model for identification of CSTR process is not yet available. As CSTR process includes tempora...
Directory of Open Access Journals (Sweden)
H Francis Song
2016-02-01
Full Text Available The ability to simultaneously record from large numbers of neurons in behaving animals has ushered in a new era for the study of the neural circuit mechanisms underlying cognitive functions. One promising approach to uncovering the dynamical and computational principles governing population responses is to analyze model recurrent neural networks (RNNs that have been optimized to perform the same tasks as behaving animals. Because the optimization of network parameters specifies the desired output but not the manner in which to achieve this output, "trained" networks serve as a source of mechanistic hypotheses and a testing ground for data analyses that link neural computation to behavior. Complete access to the activity and connectivity of the circuit, and the ability to manipulate them arbitrarily, make trained networks a convenient proxy for biological circuits and a valuable platform for theoretical investigation. However, existing RNNs lack basic biological features such as the distinction between excitatory and inhibitory units (Dale's principle, which are essential if RNNs are to provide insights into the operation of biological circuits. Moreover, trained networks can achieve the same behavioral performance but differ substantially in their structure and dynamics, highlighting the need for a simple and flexible framework for the exploratory training of RNNs. Here, we describe a framework for gradient descent-based training of excitatory-inhibitory RNNs that can incorporate a variety of biological knowledge. We provide an implementation based on the machine learning library Theano, whose automatic differentiation capabilities facilitate modifications and extensions. We validate this framework by applying it to well-known experimental paradigms such as perceptual decision-making, context-dependent integration, multisensory integration, parametric working memory, and motor sequence generation. Our results demonstrate the wide range of neural
Hoellinger, Thomas; Petieau, Mathieu; Duvinage, Matthieu; Castermans, Thierry; Seetharaman, Karthik; Cebolla, Ana-Maria; Bengoetxea, Ana; Ivanenko, Yuri; Dan, Bernard; Cheron, Guy
2013-01-01
The existence of dedicated neuronal modules such as those organized in the cerebral cortex, thalamus, basal ganglia, cerebellum, or spinal cord raises the question of how these functional modules are coordinated for appropriate motor behavior. Study of human locomotion offers an interesting field for addressing this central question. The coordination of the elevation of the 3 leg segments under a planar covariation rule (Borghese et al., 1996) was recently modeled (Barliya et al., 2009) by phase-adjusted simple oscillators shedding new light on the understanding of the central pattern generator (CPG) processing relevant oscillation signals. We describe the use of a dynamic recurrent neural network (DRNN) mimicking the natural oscillatory behavior of human locomotion for reproducing the planar covariation rule in both legs at different walking speeds. Neural network learning was based on sinusoid signals integrating frequency and amplitude features of the first three harmonics of the sagittal elevation angles of the thigh, shank, and foot of each lower limb. We verified the biological plausibility of the neural networks. Best results were obtained with oscillations extracted from the first three harmonics in comparison to oscillations outside the harmonic frequency peaks. Physiological replication steadily increased with the number of neuronal units from 1 to 80, where similarity index reached 0.99. Analysis of synaptic weighting showed that the proportion of inhibitory connections consistently increased with the number of neuronal units in the DRNN. This emerging property in the artificial neural networks resonates with recent advances in neurophysiology of inhibitory neurons that are involved in central nervous system oscillatory activities. The main message of this study is that this type of DRNN may offer a useful model of physiological central pattern generator for gaining insights in basic research and developing clinical applications.
Criticality meets learning: Criticality signatures in a self-organizing recurrent neural network.
Del Papa, Bruno; Priesemann, Viola; Triesch, Jochen
2017-01-01
Many experiments have suggested that the brain operates close to a critical state, based on signatures of criticality such as power-law distributed neuronal avalanches. In neural network models, criticality is a dynamical state that maximizes information processing capacities, e.g. sensitivity to input, dynamical range and storage capacity, which makes it a favorable candidate state for brain function. Although models that self-organize towards a critical state have been proposed, the relation between criticality signatures and learning is still unclear. Here, we investigate signatures of criticality in a self-organizing recurrent neural network (SORN). Investigating criticality in the SORN is of particular interest because it has not been developed to show criticality. Instead, the SORN has been shown to exhibit spatio-temporal pattern learning through a combination of neural plasticity mechanisms and it reproduces a number of biological findings on neural variability and the statistics and fluctuations of synaptic efficacies. We show that, after a transient, the SORN spontaneously self-organizes into a dynamical state that shows criticality signatures comparable to those found in experiments. The plasticity mechanisms are necessary to attain that dynamical state, but not to maintain it. Furthermore, onset of external input transiently changes the slope of the avalanche distributions - matching recent experimental findings. Interestingly, the membrane noise level necessary for the occurrence of the criticality signatures reduces the model's performance in simple learning tasks. Overall, our work shows that the biologically inspired plasticity and homeostasis mechanisms responsible for the SORN's spatio-temporal learning abilities can give rise to criticality signatures in its activity when driven by random input, but these break down under the structured input of short repeating sequences.
Wang, Xiao-Jing
2016-01-01
The ability to simultaneously record from large numbers of neurons in behaving animals has ushered in a new era for the study of the neural circuit mechanisms underlying cognitive functions. One promising approach to uncovering the dynamical and computational principles governing population responses is to analyze model recurrent neural networks (RNNs) that have been optimized to perform the same tasks as behaving animals. Because the optimization of network parameters specifies the desired output but not the manner in which to achieve this output, “trained” networks serve as a source of mechanistic hypotheses and a testing ground for data analyses that link neural computation to behavior. Complete access to the activity and connectivity of the circuit, and the ability to manipulate them arbitrarily, make trained networks a convenient proxy for biological circuits and a valuable platform for theoretical investigation. However, existing RNNs lack basic biological features such as the distinction between excitatory and inhibitory units (Dale’s principle), which are essential if RNNs are to provide insights into the operation of biological circuits. Moreover, trained networks can achieve the same behavioral performance but differ substantially in their structure and dynamics, highlighting the need for a simple and flexible framework for the exploratory training of RNNs. Here, we describe a framework for gradient descent-based training of excitatory-inhibitory RNNs that can incorporate a variety of biological knowledge. We provide an implementation based on the machine learning library Theano, whose automatic differentiation capabilities facilitate modifications and extensions. We validate this framework by applying it to well-known experimental paradigms such as perceptual decision-making, context-dependent integration, multisensory integration, parametric working memory, and motor sequence generation. Our results demonstrate the wide range of neural activity
Super-linear Precision in Simple Neural Population Codes
Schwab, David; Fiete, Ila
2015-03-01
A widely used tool for quantifying the precision with which a population of noisy sensory neurons encodes the value of an external stimulus is the Fisher Information (FI). Maximizing the FI is also a commonly used objective for constructing optimal neural codes. The primary utility and importance of the FI arises because it gives, through the Cramer-Rao bound, the smallest mean-squared error achievable by any unbiased stimulus estimator. However, it is well-known that when neural firing is sparse, optimizing the FI can result in codes that perform very poorly when considering the resulting mean-squared error, a measure with direct biological relevance. Here we construct optimal population codes by minimizing mean-squared error directly and study the scaling properties of the resulting network, focusing on the optimal tuning curve width. We then extend our results to continuous attractor networks that maintain short-term memory of external stimuli in their dynamics. Here we find similar scaling properties in the structure of the interactions that minimize diffusive information loss.
Quasi-projective synchronization of fractional-order complex-valued recurrent neural networks.
Yang, Shuai; Yu, Juan; Hu, Cheng; Jiang, Haijun
2018-08-01
In this paper, without separating the complex-valued neural networks into two real-valued systems, the quasi-projective synchronization of fractional-order complex-valued neural networks is investigated. First, two new fractional-order inequalities are established by using the theory of complex functions, Laplace transform and Mittag-Leffler functions, which generalize traditional inequalities with the first-order derivative in the real domain. Additionally, different from hybrid control schemes given in the previous work concerning the projective synchronization, a simple and linear control strategy is designed in this paper and several criteria are derived to ensure quasi-projective synchronization of the complex-valued neural networks with fractional-order based on the established fractional-order inequalities and the theory of complex functions. Moreover, the error bounds of quasi-projective synchronization are estimated. Especially, some conditions are also presented for the Mittag-Leffler synchronization of the addressed neural networks. Finally, some numerical examples with simulations are provided to show the effectiveness of the derived theoretical results. Copyright © 2018 Elsevier Ltd. All rights reserved.
Automatic temporal segment detection via bilateral long short-term memory recurrent neural networks
Sun, Bo; Cao, Siming; He, Jun; Yu, Lejun; Li, Liandong
2017-03-01
Constrained by the physiology, the temporal factors associated with human behavior, irrespective of facial movement or body gesture, are described by four phases: neutral, onset, apex, and offset. Although they may benefit related recognition tasks, it is not easy to accurately detect such temporal segments. An automatic temporal segment detection framework using bilateral long short-term memory recurrent neural networks (BLSTM-RNN) to learn high-level temporal-spatial features, which synthesizes the local and global temporal-spatial information more efficiently, is presented. The framework is evaluated in detail over the face and body database (FABO). The comparison shows that the proposed framework outperforms state-of-the-art methods for solving the problem of temporal segment detection.
Mizusaki, Beatriz E. P.; Agnes, Everton J.; Erichsen, Rubem; Brunnet, Leonardo G.
2017-08-01
The plastic character of brain synapses is considered to be one of the foundations for the formation of memories. There are numerous kinds of such phenomenon currently described in the literature, but their role in the development of information pathways in neural networks with recurrent architectures is still not completely clear. In this paper we study the role of an activity-based process, called pre-synaptic dependent homeostatic scaling, in the organization of networks that yield precise-timed spiking patterns. It encodes spatio-temporal information in the synaptic weights as it associates a learned input with a specific response. We introduce a correlation measure to evaluate the precision of the spiking patterns and explore the effects of different inhibitory interactions and learning parameters. We find that large learning periods are important in order to improve the network learning capacity and discuss this ability in the presence of distinct inhibitory currents.
Shahnazian, Danesh; Holroyd, Clay B
2018-02-01
Anterior cingulate cortex (ACC) has been the subject of intense debate over the past 2 decades, but its specific computational function remains controversial. Here we present a simple computational model of ACC that incorporates distributed representations across a network of interconnected processing units. Based on the proposal that ACC is concerned with the execution of extended, goal-directed action sequences, we trained a recurrent neural network to predict each successive step of several sequences associated with multiple tasks. In keeping with neurophysiological observations from nonhuman animals, the network yields distributed patterns of activity across ACC neurons that track the progression of each sequence, and in keeping with human neuroimaging data, the network produces discrepancy signals when any step of the sequence deviates from the predicted step. These simulations illustrate a novel approach for investigating ACC function.
Identification of Jets Containing $b$-Hadrons with Recurrent Neural Networks at the ATLAS Experiment
The ATLAS collaboration
2017-01-01
A novel $b$-jet identification algorithm is constructed with a Recurrent Neural Network (RNN) at the ATLAS experiment at the CERN Large Hadron Collider. The RNN based $b$-tagging algorithm processes charged particle tracks associated to jets without reliance on secondary vertex finding, and can augment existing secondary-vertex based taggers. In contrast to traditional impact-parameter-based $b$-tagging algorithms which assume that tracks associated to jets are independent from each other, the RNN based $b$-tagging algorithm can exploit the spatial and kinematic correlations between tracks which are initiated from the same $b$-hadrons. This new approach also accommodates an extended set of input variables. This note presents the expected performance of the RNN based $b$-tagging algorithm in simulated $t \\bar t$ events at $\\sqrt{s}=13$ TeV.
Han, Seong-Ik; Lee, Jang-Myung
2014-01-01
This paper proposes a backstepping control system that uses a tracking error constraint and recurrent fuzzy neural networks (RFNNs) to achieve a prescribed tracking performance for a strict-feedback nonlinear dynamic system. A new constraint variable was defined to generate the virtual control that forces the tracking error to fall within prescribed boundaries. An adaptive RFNN was also used to obtain the required improvement on the approximation performances in order to avoid calculating the explosive number of terms generated by the recursive steps of traditional backstepping control. The boundedness and convergence of the closed-loop system was confirmed based on the Lyapunov stability theory. The prescribed performance of the proposed control scheme was validated by using it to control the prescribed error of a nonlinear system and a robot manipulator. © 2013 ISA. Published by Elsevier Ltd. All rights reserved.
An Incremental Time-delay Neural Network for Dynamical Recurrent Associative Memory
Institute of Scientific and Technical Information of China (English)
无
2002-01-01
An incremental time-delay neural network based on synapse growth, which is suitable for dynamic control and learning of autonomous robots, is proposed to improve the learning and retrieving performance of dynamical recurrent associative memory architecture. The model allows steady and continuous establishment of associative memory for spatio-temporal regularities and time series in discrete sequence of inputs. The inserted hidden units can be taken as the long-term memories that expand the capacity of network and sometimes may fade away under certain condition. Preliminary experiment has shown that this incremental network may be a promising approach to endow autonomous robots with the ability of adapting to new data without destroying the learned patterns. The system also benefits from its potential chaos character for emergence.
A statistical framework for evaluating neural networks to predict recurrent events in breast cancer
Gorunescu, Florin; Gorunescu, Marina; El-Darzi, Elia; Gorunescu, Smaranda
2010-07-01
Breast cancer is the second leading cause of cancer deaths in women today. Sometimes, breast cancer can return after primary treatment. A medical diagnosis of recurrent cancer is often a more challenging task than the initial one. In this paper, we investigate the potential contribution of neural networks (NNs) to support health professionals in diagnosing such events. The NN algorithms are tested and applied to two different datasets. An extensive statistical analysis has been performed to verify our experiments. The results show that a simple network structure for both the multi-layer perceptron and radial basis function can produce equally good results, not all attributes are needed to train these algorithms and, finally, the classification performances of all algorithms are statistically robust. Moreover, we have shown that the best performing algorithm will strongly depend on the features of the datasets, and hence, there is not necessarily a single best classifier.
International Nuclear Information System (INIS)
Kim, Han Me; Kim, Jong Shik; Han, Seong Ik
2009-01-01
To improve position tracking performance of servo systems, a position tracking control using adaptive back-stepping control(ABSC) scheme and recurrent fuzzy neural networks(RFNN) is proposed. An adaptive rule of the ABSC based on system dynamics and dynamic friction model is also suggested to compensate nonlinear dynamic friction characteristics. However, it is difficult to reduce the position tracking error of servo systems by using only the ABSC scheme because of the system uncertainties which cannot be exactly identified during the modeling of servo systems. Therefore, in order to overcome system uncertainties and then to improve position tracking performance of servo systems, the RFNN technique is additionally applied to the servo system. The feasibility of the proposed control scheme for a servo system is validated through experiments. Experimental results show that the servo system with ABS controller based on the dual friction observer and RFNN including the reconstruction error estimator can achieve desired tracking performance and robustness
Jones, Kenneth Lyons; Robinson, Luther K; Benirschke, Kurt
2006-09-01
Amniotic bands can cause disruption of the cranial end of the developing fetus, leading in some cases to a neural tube closure defect. Although recurrence for unaffected parents of an affected child with a defect in which the neural tube closed normally but was subsequently disrupted by amniotic bands is negligible; for a primary defect in closure of the neural tube to which amnion has subsequently adhered, recurrence risk is 1.7%. In that primary defects of neural tube closure are characterized by typical abnormalities of the base of the skull, evaluation of the cranial base in such fetuses provides an approach for making a distinction between these 2 mechanisms. This distinction has implications regarding recurrence risk. The skull base of 2 fetuses with amnion rupture sequence involving the cranial end of the neural tube were compared to that of 1 fetus with anencephaly as well as that of a structurally normal fetus. The skulls were cleaned, fixed in 10% formalin, recleaned, and then exposed to 10% KOH solution. After washing and recleaning, the skulls were exposed to hydrogen peroxide for bleaching and photography. Despite involvement of the anterior neural tube in both fetuses with amnion rupture sequence, in Case 3 the cranial base was normal while in Case 4 the cranial base was similar to that seen in anencephaly. This technique provides a method for determining the developmental pathogenesis of anterior neural tube defects in cases of amnion rupture sequence. As such, it provides information that can be used to counsel parents of affected children with respect to recurrence risk.
Using recurrent neural network models for early detection of heart failure onset.
Choi, Edward; Schuetz, Andy; Stewart, Walter F; Sun, Jimeng
2017-03-01
We explored whether use of deep learning to model temporal relations among events in electronic health records (EHRs) would improve model performance in predicting initial diagnosis of heart failure (HF) compared to conventional methods that ignore temporality. Data were from a health system's EHR on 3884 incident HF cases and 28 903 controls, identified as primary care patients, between May 16, 2000, and May 23, 2013. Recurrent neural network (RNN) models using gated recurrent units (GRUs) were adapted to detect relations among time-stamped events (eg, disease diagnosis, medication orders, procedure orders, etc.) with a 12- to 18-month observation window of cases and controls. Model performance metrics were compared to regularized logistic regression, neural network, support vector machine, and K-nearest neighbor classifier approaches. Using a 12-month observation window, the area under the curve (AUC) for the RNN model was 0.777, compared to AUCs for logistic regression (0.747), multilayer perceptron (MLP) with 1 hidden layer (0.765), support vector machine (SVM) (0.743), and K-nearest neighbor (KNN) (0.730). When using an 18-month observation window, the AUC for the RNN model increased to 0.883 and was significantly higher than the 0.834 AUC for the best of the baseline methods (MLP). Deep learning models adapted to leverage temporal relations appear to improve performance of models for detection of incident heart failure with a short observation window of 12-18 months. © The Author 2016. Published by Oxford University Press on behalf of the American Medical Informatics Association.
Design of a heart rate controller for treadmill exercise using a recurrent fuzzy neural network.
Lu, Chun-Hao; Wang, Wei-Cheng; Tai, Cheng-Chi; Chen, Tien-Chi
2016-05-01
In this study, we developed a computer controlled treadmill system using a recurrent fuzzy neural network heart rate controller (RFNNHRC). Treadmill speeds and inclines were controlled by corresponding control servo motors. The RFNNHRC was used to generate the control signals to automatically control treadmill speed and incline to minimize the user heart rate deviations from a preset profile. The RFNNHRC combines a fuzzy reasoning capability to accommodate uncertain information and an artificial recurrent neural network learning process that corrects for treadmill system nonlinearities and uncertainties. Treadmill speeds and inclines are controlled by the RFNNHRC to achieve minimal heart rate deviation from a pre-set profile using adjustable parameters and an on-line learning algorithm that provides robust performance against parameter variations. The on-line learning algorithm of RFNNHRC was developed and implemented using a dsPIC 30F4011 DSP. Application of the proposed control scheme to heart rate responses of runners resulted in smaller fluctuations than those produced by using proportional integra control, and treadmill speeds and inclines were smoother. The present experiments demonstrate improved heart rate tracking performance with the proposed control scheme. The RFNNHRC scheme with adjustable parameters and an on-line learning algorithm was applied to a computer controlled treadmill system with heart rate control during treadmill exercise. Novel RFNNHRC structure and controller stability analyses were introduced. The RFNNHRC were tuned using a Lyapunov function to ensure system stability. The superior heart rate control with the proposed RFNNHRC scheme was demonstrated with various pre-set heart rates. Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.
Neural processing of short-term recurrence in songbird vocal communication.
Directory of Open Access Journals (Sweden)
Gabriël J L Beckers
Full Text Available BACKGROUND: Many situations involving animal communication are dominated by recurring, stereotyped signals. How do receivers optimally distinguish between frequently recurring signals and novel ones? Cortical auditory systems are known to be pre-attentively sensitive to short-term delivery statistics of artificial stimuli, but it is unknown if this phenomenon extends to the level of behaviorally relevant delivery patterns, such as those used during communication. METHODOLOGY/PRINCIPAL FINDINGS: We recorded and analyzed complete auditory scenes of spontaneously communicating zebra finch (Taeniopygia guttata pairs over a week-long period, and show that they can produce tens of thousands of short-range contact calls per day. Individual calls recur at time scales (median interval 1.5 s matching those at which mammalian sensory systems are sensitive to recent stimulus history. Next, we presented to anesthetized birds sequences of frequently recurring calls interspersed with rare ones, and recorded, in parallel, action and local field potential responses in the medio-caudal auditory forebrain at 32 unique sites. Variation in call recurrence rate over natural ranges leads to widespread and significant modulation in strength of neural responses. Such modulation is highly call-specific in secondary auditory areas, but not in the main thalamo-recipient, primary auditory area. CONCLUSIONS/SIGNIFICANCE: Our results support the hypothesis that pre-attentive neural sensitivity to short-term stimulus recurrence is involved in the analysis of auditory scenes at the level of delivery patterns of meaningful sounds. This may enable birds to efficiently and automatically distinguish frequently recurring vocalizations from other events in their auditory scene.
A new neural network model for solving random interval linear programming problems.
Arjmandzadeh, Ziba; Safi, Mohammadreza; Nazemi, Alireza
2017-05-01
This paper presents a neural network model for solving random interval linear programming problems. The original problem involving random interval variable coefficients is first transformed into an equivalent convex second order cone programming problem. A neural network model is then constructed for solving the obtained convex second order cone problem. Employing Lyapunov function approach, it is also shown that the proposed neural network model is stable in the sense of Lyapunov and it is globally convergent to an exact satisfactory solution of the original problem. Several illustrative examples are solved in support of this technique. Copyright © 2017 Elsevier Ltd. All rights reserved.
Linear and nonlinear ARMA model parameter estimation using an artificial neural network
Chon, K. H.; Cohen, R. J.
1997-01-01
This paper addresses parametric system identification of linear and nonlinear dynamic systems by analysis of the input and output signals. Specifically, we investigate the relationship between estimation of the system using a feedforward neural network model and estimation of the system by use of linear and nonlinear autoregressive moving-average (ARMA) models. By utilizing a neural network model incorporating a polynomial activation function, we show the equivalence of the artificial neural network to the linear and nonlinear ARMA models. We compare the parameterization of the estimated system using the neural network and ARMA approaches by utilizing data generated by means of computer simulations. Specifically, we show that the parameters of a simulated ARMA system can be obtained from the neural network analysis of the simulated data or by conventional least squares ARMA analysis. The feasibility of applying neural networks with polynomial activation functions to the analysis of experimental data is explored by application to measurements of heart rate (HR) and instantaneous lung volume (ILV) fluctuations.
Wu, Wei; Cui, Bao-Tong
2007-07-01
In this paper, a synchronization scheme for a class of chaotic neural networks with time-varying delays is presented. This class of chaotic neural networks covers several well-known neural networks, such as Hopfield neural networks, cellular neural networks, and bidirectional associative memory networks. The obtained criteria are expressed in terms of linear matrix inequalities, thus they can be efficiently verified. A comparison between our results and the previous results shows that our results are less restrictive.
Gelderloos, L.J.; Chrupala, Grzegorz
2016-01-01
We present a model of visually-grounded language learning based on stacked gated recurrent neural networks which learns to predict visual features given an image description in the form of a sequence of phonemes. The learning task resembles that faced by human language learners who need to discover
Sequence-specific bias correction for RNA-seq data using recurrent neural networks.
Zhang, Yao-Zhong; Yamaguchi, Rui; Imoto, Seiya; Miyano, Satoru
2017-01-25
The recent success of deep learning techniques in machine learning and artificial intelligence has stimulated a great deal of interest among bioinformaticians, who now wish to bring the power of deep learning to bare on a host of bioinformatical problems. Deep learning is ideally suited for biological problems that require automatic or hierarchical feature representation for biological data when prior knowledge is limited. In this work, we address the sequence-specific bias correction problem for RNA-seq data redusing Recurrent Neural Networks (RNNs) to model nucleotide sequences without pre-determining sequence structures. The sequence-specific bias of a read is then calculated based on the sequence probabilities estimated by RNNs, and used in the estimation of gene abundance. We explore the application of two popular RNN recurrent units for this task and demonstrate that RNN-based approaches provide a flexible way to model nucleotide sequences without knowledge of predetermined sequence structures. Our experiments show that training a RNN-based nucleotide sequence model is efficient and RNN-based bias correction methods compare well with the-state-of-the-art sequence-specific bias correction method on the commonly used MAQC-III data set. RNNs provides an alternative and flexible way to calculate sequence-specific bias without explicitly pre-determining sequence structures.
Marginally Stable Triangular Recurrent Neural Network Architecture for Time Series Prediction.
Sivakumar, Seshadri; Sivakumar, Shyamala
2017-09-25
This paper introduces a discrete-time recurrent neural network architecture using triangular feedback weight matrices that allows a simplified approach to ensuring network and training stability. The triangular structure of the weight matrices is exploited to readily ensure that the eigenvalues of the feedback weight matrix represented by the block diagonal elements lie on the unit circle in the complex z-plane by updating these weights based on the differential of the angular error variable. Such placement of the eigenvalues together with the extended close interaction between state variables facilitated by the nondiagonal triangular elements, enhances the learning ability of the proposed architecture. Simulation results show that the proposed architecture is highly effective in time-series prediction tasks associated with nonlinear and chaotic dynamic systems with underlying oscillatory modes. This modular architecture with dual upper and lower triangular feedback weight matrices mimics fully recurrent network architectures, while maintaining learning stability with a simplified training process. While training, the block-diagonal weights (hence the eigenvalues) of the dual triangular matrices are constrained to the same values during weight updates aimed at minimizing the possibility of overfitting. The dual triangular architecture also exploits the benefit of parsing the input and selectively applying the parsed inputs to the two subnetworks to facilitate enhanced learning performance.
Wang, Yunlong; Liu, Fei; Zhang, Kunbo; Hou, Guangqi; Sun, Zhenan; Tan, Tieniu
2018-09-01
The low spatial resolution of light-field image poses significant difficulties in exploiting its advantage. To mitigate the dependency of accurate depth or disparity information as priors for light-field image super-resolution, we propose an implicitly multi-scale fusion scheme to accumulate contextual information from multiple scales for super-resolution reconstruction. The implicitly multi-scale fusion scheme is then incorporated into bidirectional recurrent convolutional neural network, which aims to iteratively model spatial relations between horizontally or vertically adjacent sub-aperture images of light-field data. Within the network, the recurrent convolutions are modified to be more effective and flexible in modeling the spatial correlations between neighboring views. A horizontal sub-network and a vertical sub-network of the same network structure are ensembled for final outputs via stacked generalization. Experimental results on synthetic and real-world data sets demonstrate that the proposed method outperforms other state-of-the-art methods by a large margin in peak signal-to-noise ratio and gray-scale structural similarity indexes, which also achieves superior quality for human visual systems. Furthermore, the proposed method can enhance the performance of light field applications such as depth estimation.
Deep Recurrent Neural Networks for seizure detection and early seizure detection systems
Energy Technology Data Exchange (ETDEWEB)
Talathi, S. S. [Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
2017-06-05
Epilepsy is common neurological diseases, affecting about 0.6-0.8 % of world population. Epileptic patients suffer from chronic unprovoked seizures, which can result in broad spectrum of debilitating medical and social consequences. Since seizures, in general, occur infrequently and are unpredictable, automated seizure detection systems are recommended to screen for seizures during long-term electroencephalogram (EEG) recordings. In addition, systems for early seizure detection can lead to the development of new types of intervention systems that are designed to control or shorten the duration of seizure events. In this article, we investigate the utility of recurrent neural networks (RNNs) in designing seizure detection and early seizure detection systems. We propose a deep learning framework via the use of Gated Recurrent Unit (GRU) RNNs for seizure detection. We use publicly available data in order to evaluate our method and demonstrate very promising evaluation results with overall accuracy close to 100 %. We also systematically investigate the application of our method for early seizure warning systems. Our method can detect about 98% of seizure events within the first 5 seconds of the overall epileptic seizure duration.
Raghu, S; Sriraam, N; Kumar, G Pradeep
2017-02-01
Electroencephalogram shortly termed as EEG is considered as the fundamental segment for the assessment of the neural activities in the brain. In cognitive neuroscience domain, EEG-based assessment method is found to be superior due to its non-invasive ability to detect deep brain structure while exhibiting superior spatial resolutions. Especially for studying the neurodynamic behavior of epileptic seizures, EEG recordings reflect the neuronal activity of the brain and thus provide required clinical diagnostic information for the neurologist. This specific proposed study makes use of wavelet packet based log and norm entropies with a recurrent Elman neural network (REN) for the automated detection of epileptic seizures. Three conditions, normal, pre-ictal and epileptic EEG recordings were considered for the proposed study. An adaptive Weiner filter was initially applied to remove the power line noise of 50 Hz from raw EEG recordings. Raw EEGs were segmented into 1 s patterns to ensure stationarity of the signal. Then wavelet packet using Haar wavelet with a five level decomposition was introduced and two entropies, log and norm were estimated and were applied to REN classifier to perform binary classification. The non-linear Wilcoxon statistical test was applied to observe the variation in the features under these conditions. The effect of log energy entropy (without wavelets) was also studied. It was found from the simulation results that the wavelet packet log entropy with REN classifier yielded a classification accuracy of 99.70 % for normal-pre-ictal, 99.70 % for normal-epileptic and 99.85 % for pre-ictal-epileptic.
Zazo, Ruben; Lozano-Diez, Alicia; Gonzalez-Dominguez, Javier; Toledano, Doroteo T; Gonzalez-Rodriguez, Joaquin
2016-01-01
Long Short Term Memory (LSTM) Recurrent Neural Networks (RNNs) have recently outperformed other state-of-the-art approaches, such as i-vector and Deep Neural Networks (DNNs), in automatic Language Identification (LID), particularly when dealing with very short utterances (∼3s). In this contribution we present an open-source, end-to-end, LSTM RNN system running on limited computational resources (a single GPU) that outperforms a reference i-vector system on a subset of the NIST Language Recognition Evaluation (8 target languages, 3s task) by up to a 26%. This result is in line with previously published research using proprietary LSTM implementations and huge computational resources, which made these former results hardly reproducible. Further, we extend those previous experiments modeling unseen languages (out of set, OOS, modeling), which is crucial in real applications. Results show that a LSTM RNN with OOS modeling is able to detect these languages and generalizes robustly to unseen OOS languages. Finally, we also analyze the effect of even more limited test data (from 2.25s to 0.1s) proving that with as little as 0.5s an accuracy of over 50% can be achieved.
Recurrent neural networks for breast lesion classification based on DCE-MRIs
Antropova, Natasha; Huynh, Benjamin; Giger, Maryellen
2018-02-01
Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) plays a significant role in breast cancer screening, cancer staging, and monitoring response to therapy. Recently, deep learning methods are being rapidly incorporated in image-based breast cancer diagnosis and prognosis. However, most of the current deep learning methods make clinical decisions based on 2-dimentional (2D) or 3D images and are not well suited for temporal image data. In this study, we develop a deep learning methodology that enables integration of clinically valuable temporal components of DCE-MRIs into deep learning-based lesion classification. Our work is performed on a database of 703 DCE-MRI cases for the task of distinguishing benign and malignant lesions, and uses the area under the ROC curve (AUC) as the performance metric in conducting that task. We train a recurrent neural network, specifically a long short-term memory network (LSTM), on sequences of image features extracted from the dynamic MRI sequences. These features are extracted with VGGNet, a convolutional neural network pre-trained on a large dataset of natural images ImageNet. The features are obtained from various levels of the network, to capture low-, mid-, and high-level information about the lesion. Compared to a classification method that takes as input only images at a single time-point (yielding an AUC = 0.81 (se = 0.04)), our LSTM method improves lesion classification with an AUC of 0.85 (se = 0.03).
Using Long-Short-Term-Memory Recurrent Neural Networks to Predict Aviation Engine Vibrations
ElSaid, AbdElRahman Ahmed
This thesis examines building viable Recurrent Neural Networks (RNN) using Long Short Term Memory (LSTM) neurons to predict aircraft engine vibrations. The different networks are trained on a large database of flight data records obtained from an airline containing flights that suffered from excessive vibration. RNNs can provide a more generalizable and robust method for prediction over analytical calculations of engine vibration, as analytical calculations must be solved iteratively based on specific empirical engine parameters, and this database contains multiple types of engines. Further, LSTM RNNs provide a "memory" of the contribution of previous time series data which can further improve predictions of future vibration values. LSTM RNNs were used over traditional RNNs, as those suffer from vanishing/exploding gradients when trained with back propagation. The study managed to predict vibration values for 1, 5, 10, and 20 seconds in the future, with 2.84% 3.3%, 5.51% and 10.19% mean absolute error, respectively. These neural networks provide a promising means for the future development of warning systems so that suitable actions can be taken before the occurrence of excess vibration to avoid unfavorable situations during flight.
Nonlinear dynamics analysis of a self-organizing recurrent neural network: chaos waning.
Eser, Jürgen; Zheng, Pengsheng; Triesch, Jochen
2014-01-01
Self-organization is thought to play an important role in structuring nervous systems. It frequently arises as a consequence of plasticity mechanisms in neural networks: connectivity determines network dynamics which in turn feed back on network structure through various forms of plasticity. Recently, self-organizing recurrent neural network models (SORNs) have been shown to learn non-trivial structure in their inputs and to reproduce the experimentally observed statistics and fluctuations of synaptic connection strengths in cortex and hippocampus. However, the dynamics in these networks and how they change with network evolution are still poorly understood. Here we investigate the degree of chaos in SORNs by studying how the networks' self-organization changes their response to small perturbations. We study the effect of perturbations to the excitatory-to-excitatory weight matrix on connection strengths and on unit activities. We find that the network dynamics, characterized by an estimate of the maximum Lyapunov exponent, becomes less chaotic during its self-organization, developing into a regime where only few perturbations become amplified. We also find that due to the mixing of discrete and (quasi-)continuous variables in SORNs, small perturbations to the synaptic weights may become amplified only after a substantial delay, a phenomenon we propose to call deferred chaos.
International Nuclear Information System (INIS)
Sabahi, Kamel; Teshnehlab, Mohammad; Shoorhedeli, Mahdi Aliyari
2009-01-01
In this study, a new adaptive controller based on modified feedback error learning (FEL) approaches is proposed for load frequency control (LFC) problem. The FEL strategy consists of intelligent and conventional controllers in feedforward and feedback paths, respectively. In this strategy, a conventional feedback controller (CFC), i.e. proportional, integral and derivative (PID) controller, is essential to guarantee global asymptotic stability of the overall system; and an intelligent feedforward controller (INFC) is adopted to learn the inverse of the controlled system. Therefore, when the INFC learns the inverse of controlled system, the tracking of reference signal is done properly. Generally, the CFC is designed at nominal operating conditions of the system and, therefore, fails to provide the best control performance as well as global stability over a wide range of changes in the operating conditions of the system. So, in this study a supervised controller (SC), a lookup table based controller, is addressed for tuning of the CFC. During abrupt changes of the power system parameters, the SC adjusts the PID parameters according to these operating conditions. Moreover, for improving the performance of overall system, a recurrent fuzzy neural network (RFNN) is adopted in INFC instead of the conventional neural network, which was used in past studies. The proposed FEL controller has been compared with the conventional feedback error learning controller (CFEL) and the PID controller through some performance indices
A recurrent neural model for proto-object based contour integration and figure-ground segregation.
Hu, Brian; Niebur, Ernst
2017-12-01
Visual processing of objects makes use of both feedforward and feedback streams of information. However, the nature of feedback signals is largely unknown, as is the identity of the neuronal populations in lower visual areas that receive them. Here, we develop a recurrent neural model to address these questions in the context of contour integration and figure-ground segregation. A key feature of our model is the use of grouping neurons whose activity represents tentative objects ("proto-objects") based on the integration of local feature information. Grouping neurons receive input from an organized set of local feature neurons, and project modulatory feedback to those same neurons. Additionally, inhibition at both the local feature level and the object representation level biases the interpretation of the visual scene in agreement with principles from Gestalt psychology. Our model explains several sets of neurophysiological results (Zhou et al. Journal of Neuroscience, 20(17), 6594-6611 2000; Qiu et al. Nature Neuroscience, 10(11), 1492-1499 2007; Chen et al. Neuron, 82(3), 682-694 2014), and makes testable predictions about the influence of neuronal feedback and attentional selection on neural responses across different visual areas. Our model also provides a framework for understanding how object-based attention is able to select both objects and the features associated with them.
International Nuclear Information System (INIS)
Zhang Yunong; Li Zhan
2009-01-01
In this Letter, by following Zhang et al.'s method, a recurrent neural network (termed as Zhang neural network, ZNN) is developed and analyzed for solving online the time-varying convex quadratic-programming problem subject to time-varying linear-equality constraints. Different from conventional gradient-based neural networks (GNN), such a ZNN model makes full use of the time-derivative information of time-varying coefficient. The resultant ZNN model is theoretically proved to have global exponential convergence to the time-varying theoretical optimal solution of the investigated time-varying convex quadratic program. Computer-simulation results further substantiate the effectiveness, efficiency and novelty of such ZNN model and method.
Hwang, Chih-Lyang; Jan, Chau
2016-02-01
At the beginning, an approximate nonlinear autoregressive moving average (NARMA) model is employed to represent a class of multivariable nonlinear dynamic systems with time-varying delay. It is known that the disadvantages of robust control for the NARMA model are as follows: 1) suitable control parameters for larger time delay are more sensitive to achieving desirable performance; 2) it only deals with bounded uncertainty; and 3) the nominal NARMA model must be learned in advance. Due to the dynamic feature of the NARMA model, a recurrent neural network (RNN) is online applied to learn it. However, the system performance becomes deteriorated due to the poor learning of the larger variation of system vector functions. In this situation, a simple network is employed to compensate the upper bound of the residue caused by the linear parameterization of the approximation error of RNN. An e -modification learning law with a projection for weight matrix is applied to guarantee its boundedness without persistent excitation. Under suitable conditions, the semiglobally ultimately bounded tracking with the boundedness of estimated weight matrix is obtained by the proposed RNN-based multivariable adaptive control. Finally, simulations are presented to verify the effectiveness and robustness of the proposed control.
Predicting musically induced emotions from physiological inputs: linear and neural network models.
Russo, Frank A; Vempala, Naresh N; Sandstrom, Gillian M
2013-01-01
Listening to music often leads to physiological responses. Do these physiological responses contain sufficient information to infer emotion induced in the listener? The current study explores this question by attempting to predict judgments of "felt" emotion from physiological responses alone using linear and neural network models. We measured five channels of peripheral physiology from 20 participants-heart rate (HR), respiration, galvanic skin response, and activity in corrugator supercilii and zygomaticus major facial muscles. Using valence and arousal (VA) dimensions, participants rated their felt emotion after listening to each of 12 classical music excerpts. After extracting features from the five channels, we examined their correlation with VA ratings, and then performed multiple linear regression to see if a linear relationship between the physiological responses could account for the ratings. Although linear models predicted a significant amount of variance in arousal ratings, they were unable to do so with valence ratings. We then used a neural network to provide a non-linear account of the ratings. The network was trained on the mean ratings of eight of the 12 excerpts and tested on the remainder. Performance of the neural network confirms that physiological responses alone can be used to predict musically induced emotion. The non-linear model derived from the neural network was more accurate than linear models derived from multiple linear regression, particularly along the valence dimension. A secondary analysis allowed us to quantify the relative contributions of inputs to the non-linear model. The study represents a novel approach to understanding the complex relationship between physiological responses and musically induced emotion.
International Nuclear Information System (INIS)
Lu Junguo; Lu Linji
2009-01-01
In this paper, global exponential stability and periodicity of a class of reaction-diffusion recurrent neural networks with distributed delays and Dirichlet boundary conditions are studied by constructing suitable Lyapunov functionals and utilizing some inequality techniques. We first prove global exponential convergence to 0 of the difference between any two solutions of the original neural networks, the existence and uniqueness of equilibrium is the direct results of this procedure. This approach is different from the usually used one where the existence, uniqueness of equilibrium and stability are proved in two separate steps. Secondly, we prove periodicity. Sufficient conditions ensuring the existence, uniqueness, and global exponential stability of the equilibrium and periodic solution are given. These conditions are easy to verify and our results play an important role in the design and application of globally exponentially stable neural circuits and periodic oscillatory neural circuits.
How linear response shaped models of neural circuits and the quest for alternatives.
Herfurth, Tim; Tchumatchenko, Tatjana
2017-10-01
In the past decades, many mathematical approaches to solve complex nonlinear systems in physics have been successfully applied to neuroscience. One of these tools is the concept of linear response functions. However, phenomena observed in the brain emerge from fundamentally nonlinear interactions and feedback loops rather than from a composition of linear filters. Here, we review the successes achieved by applying the linear response formalism to topics, such as rhythm generation and synchrony and by incorporating it into models that combine linear and nonlinear transformations. We also discuss the challenges encountered in the linear response applications and argue that new theoretical concepts are needed to tackle feedback loops and non-equilibrium dynamics which are experimentally observed in neural networks but are outside of the validity regime of the linear response formalism. Copyright © 2017 Elsevier Ltd. All rights reserved.
Monitoring tool usage in surgery videos using boosted convolutional and recurrent neural networks.
Al Hajj, Hassan; Lamard, Mathieu; Conze, Pierre-Henri; Cochener, Béatrice; Quellec, Gwenolé
2018-05-09
This paper investigates the automatic monitoring of tool usage during a surgery, with potential applications in report generation, surgical training and real-time decision support. Two surgeries are considered: cataract surgery, the most common surgical procedure, and cholecystectomy, one of the most common digestive surgeries. Tool usage is monitored in videos recorded either through a microscope (cataract surgery) or an endoscope (cholecystectomy). Following state-of-the-art video analysis solutions, each frame of the video is analyzed by convolutional neural networks (CNNs) whose outputs are fed to recurrent neural networks (RNNs) in order to take temporal relationships between events into account. Novelty lies in the way those CNNs and RNNs are trained. Computational complexity prevents the end-to-end training of "CNN+RNN" systems. Therefore, CNNs are usually trained first, independently from the RNNs. This approach is clearly suboptimal for surgical tool analysis: many tools are very similar to one another, but they can generally be differentiated based on past events. CNNs should be trained to extract the most useful visual features in combination with the temporal context. A novel boosting strategy is proposed to achieve this goal: the CNN and RNN parts of the system are simultaneously enriched by progressively adding weak classifiers (either CNNs or RNNs) trained to improve the overall classification accuracy. Experiments were performed in a dataset of 50 cataract surgery videos, where the usage of 21 surgical tools was manually annotated, and a dataset of 80 cholecystectomy videos, where the usage of 7 tools was manually annotated. Very good classification performance are achieved in both datasets: tool usage could be labeled with an average area under the ROC curve of A z =0.9961 and A z =0.9939, respectively, in offline mode (using past, present and future information), and A z =0.9957 and A z =0.9936, respectively, in online mode (using past and present
Fast non-linear extraction of plasma equilibrium parameters using a neural network mapping
International Nuclear Information System (INIS)
Lister, J.B.; Schnurrenberger, H.
1990-07-01
The shaping of non-circular plasmas requires a non-linear mapping between the measured diagnostic signals and selected equilibrium parameters. The particular configuration of Neural Network known as the multi-layer perceptron provides a powerful and general technique for formulating an arbitrary continuous non-linear multi-dimensional mapping. This technique has been successfully applied to the extraction of equilibrium parameters from measurements of single-null diverted plasmas in the DIII-D tokamak; the results are compared with a purely linear mapping. The method is promising, and hardware implementation is straightforward. (author) 15 refs., 7 figs
Fast non-linear extraction of plasma equilibrium parameters using a neural network mapping
International Nuclear Information System (INIS)
Lister, J.B.; Schnurrenberger, H.
1991-01-01
The shaping of non-circular plasmas requires a non-linear mapping between the measured diagnostic signals and selected equilibrium parameters. The particular configuration of neural network known as the multilayer perceptron provides a powerful and general technique for formulating an arbitrary continuous non-linear multi-dimensional mapping. This technique has been successfully applied to the extraction of equilibrium parameters from measurements of single-null diverted plasmas in the DIII-D tokamak; the results are compared with a purely linear mapping. The method is promising, and hardware implementation is straightforward. (author). 17 refs, 8 figs, 2 tab
Cichocki, A; Unbehauen, R
1994-01-01
In this paper a new class of simplified low-cost analog artificial neural networks with on chip adaptive learning algorithms are proposed for solving linear systems of algebraic equations in real time. The proposed learning algorithms for linear least squares (LS), total least squares (TLS) and data least squares (DLS) problems can be considered as modifications and extensions of well known algorithms: the row-action projection-Kaczmarz algorithm and/or the LMS (Adaline) Widrow-Hoff algorithms. The algorithms can be applied to any problem which can be formulated as a linear regression problem. The correctness and high performance of the proposed neural networks are illustrated by extensive computer simulation results.
Lu, Wenlian; Zheng, Ren; Chen, Tianping
2016-03-01
In this paper, we discuss outer-synchronization of the asymmetrically connected recurrent time-varying neural networks. By using both centralized and decentralized discretization data sampling principles, we derive several sufficient conditions based on three vector norms to guarantee that the difference of any two trajectories starting from different initial values of the neural network converges to zero. The lower bounds of the common time intervals between data samples in centralized and decentralized principles are proved to be positive, which guarantees exclusion of Zeno behavior. A numerical example is provided to illustrate the efficiency of the theoretical results. Copyright © 2015 Elsevier Ltd. All rights reserved.
Predicting musically induced emotions from physiological inputs: Linear and neural network models
Directory of Open Access Journals (Sweden)
Frank A. Russo
2013-08-01
Full Text Available Listening to music often leads to physiological responses. Do these physiological responses contain sufficient information to infer emotion induced in the listener? The current study explores this question by attempting to predict judgments of 'felt' emotion from physiological responses alone using linear and neural network models. We measured five channels of peripheral physiology from 20 participants – heart rate, respiration, galvanic skin response, and activity in corrugator supercilii and zygomaticus major facial muscles. Using valence and arousal (VA dimensions, participants rated their felt emotion after listening to each of 12 classical music excerpts. After extracting features from the five channels, we examined their correlation with VA ratings, and then performed multiple linear regression to see if a linear relationship between the physiological responses could account for the ratings. Although linear models predicted a significant amount of variance in arousal ratings, they were unable to do so with valence ratings. We then used a neural network to provide a nonlinear account of the ratings. The network was trained on the mean ratings of eight of the 12 excerpts and tested on the remainder. Performance of the neural network confirms that physiological responses alone can be used to predict musically induced emotion. The nonlinear model derived from the neural network was more accurate than linear models derived from multiple linear regression, particularly along the valence dimension. A secondary analysis allowed us to quantify the relative contributions of inputs to the nonlinear model. The study represents a novel approach to understanding the complex relationship between physiological responses and musically induced emotion.
Protein Solvent-Accessibility Prediction by a Stacked Deep Bidirectional Recurrent Neural Network
Directory of Open Access Journals (Sweden)
Buzhong Zhang
2018-05-01
Full Text Available Residue solvent accessibility is closely related to the spatial arrangement and packing of residues. Predicting the solvent accessibility of a protein is an important step to understand its structure and function. In this work, we present a deep learning method to predict residue solvent accessibility, which is based on a stacked deep bidirectional recurrent neural network applied to sequence profiles. To capture more long-range sequence information, a merging operator was proposed when bidirectional information from hidden nodes was merged for outputs. Three types of merging operators were used in our improved model, with a long short-term memory network performing as a hidden computing node. The trained database was constructed from 7361 proteins extracted from the PISCES server using a cut-off of 25% sequence identity. Sequence-derived features including position-specific scoring matrix, physical properties, physicochemical characteristics, conservation score and protein coding were used to represent a residue. Using this method, predictive values of continuous relative solvent-accessible area were obtained, and then, these values were transformed into binary states with predefined thresholds. Our experimental results showed that our deep learning method improved prediction quality relative to current methods, with mean absolute error and Pearson’s correlation coefficient values of 8.8% and 74.8%, respectively, on the CB502 dataset and 8.2% and 78%, respectively, on the Manesh215 dataset.
Protein Solvent-Accessibility Prediction by a Stacked Deep Bidirectional Recurrent Neural Network.
Zhang, Buzhong; Li, Linqing; Lü, Qiang
2018-05-25
Residue solvent accessibility is closely related to the spatial arrangement and packing of residues. Predicting the solvent accessibility of a protein is an important step to understand its structure and function. In this work, we present a deep learning method to predict residue solvent accessibility, which is based on a stacked deep bidirectional recurrent neural network applied to sequence profiles. To capture more long-range sequence information, a merging operator was proposed when bidirectional information from hidden nodes was merged for outputs. Three types of merging operators were used in our improved model, with a long short-term memory network performing as a hidden computing node. The trained database was constructed from 7361 proteins extracted from the PISCES server using a cut-off of 25% sequence identity. Sequence-derived features including position-specific scoring matrix, physical properties, physicochemical characteristics, conservation score and protein coding were used to represent a residue. Using this method, predictive values of continuous relative solvent-accessible area were obtained, and then, these values were transformed into binary states with predefined thresholds. Our experimental results showed that our deep learning method improved prediction quality relative to current methods, with mean absolute error and Pearson's correlation coefficient values of 8.8% and 74.8%, respectively, on the CB502 dataset and 8.2% and 78%, respectively, on the Manesh215 dataset.
Fei, Juntao; Lu, Cheng
2018-04-01
In this paper, an adaptive sliding mode control system using a double loop recurrent neural network (DLRNN) structure is proposed for a class of nonlinear dynamic systems. A new three-layer RNN is proposed to approximate unknown dynamics with two different kinds of feedback loops where the firing weights and output signal calculated in the last step are stored and used as the feedback signals in each feedback loop. Since the new structure has combined the advantages of internal feedback NN and external feedback NN, it can acquire the internal state information while the output signal is also captured, thus the new designed DLRNN can achieve better approximation performance compared with the regular NNs without feedback loops or the regular RNNs with a single feedback loop. The new proposed DLRNN structure is employed in an equivalent controller to approximate the unknown nonlinear system dynamics, and the parameters of the DLRNN are updated online by adaptive laws to get favorable approximation performance. To investigate the effectiveness of the proposed controller, the designed adaptive sliding mode controller with the DLRNN is applied to a -axis microelectromechanical system gyroscope to control the vibrating dynamics of the proof mass. Simulation results demonstrate that the proposed methodology can achieve good tracking property, and the comparisons of the approximation performance between radial basis function NN, RNN, and DLRNN show that the DLRNN can accurately estimate the unknown dynamics with a fast speed while the internal states of DLRNN are more stable.
A recurrent neural network for classification of unevenly sampled variable stars
Naul, Brett; Bloom, Joshua S.; Pérez, Fernando; van der Walt, Stéfan
2018-02-01
Astronomical surveys of celestial sources produce streams of noisy time series measuring flux versus time (`light curves'). Unlike in many other physical domains, however, large (and source-specific) temporal gaps in data arise naturally due to intranight cadence choices as well as diurnal and seasonal constraints1-5. With nightly observations of millions of variable stars and transients from upcoming surveys4,6, efficient and accurate discovery and classification techniques on noisy, irregularly sampled data must be employed with minimal human-in-the-loop involvement. Machine learning for inference tasks on such data traditionally requires the laborious hand-coding of domain-specific numerical summaries of raw data (`features')7. Here, we present a novel unsupervised autoencoding recurrent neural network8 that makes explicit use of sampling times and known heteroskedastic noise properties. When trained on optical variable star catalogues, this network produces supervised classification models that rival other best-in-class approaches. We find that autoencoded features learned in one time-domain survey perform nearly as well when applied to another survey. These networks can continue to learn from new unlabelled observations and may be used in other unsupervised tasks, such as forecasting and anomaly detection.
Mandal, Sudip; Saha, Goutam; Pal, Rajat Kumar
2017-08-01
Correct inference of genetic regulations inside a cell from the biological database like time series microarray data is one of the greatest challenges in post genomic era for biologists and researchers. Recurrent Neural Network (RNN) is one of the most popular and simple approach to model the dynamics as well as to infer correct dependencies among genes. Inspired by the behavior of social elephants, we propose a new metaheuristic namely Elephant Swarm Water Search Algorithm (ESWSA) to infer Gene Regulatory Network (GRN). This algorithm is mainly based on the water search strategy of intelligent and social elephants during drought, utilizing the different types of communication techniques. Initially, the algorithm is tested against benchmark small and medium scale artificial genetic networks without and with presence of different noise levels and the efficiency was observed in term of parametric error, minimum fitness value, execution time, accuracy of prediction of true regulation, etc. Next, the proposed algorithm is tested against the real time gene expression data of Escherichia Coli SOS Network and results were also compared with others state of the art optimization methods. The experimental results suggest that ESWSA is very efficient for GRN inference problem and performs better than other methods in many ways.
Applying long short-term memory recurrent neural networks to intrusion detection
Directory of Open Access Journals (Sweden)
Ralf C. Staudemeyer
2015-07-01
Full Text Available We claim that modelling network traffic as a time series with a supervised learning approach, using known genuine and malicious behaviour, improves intrusion detection. To substantiate this, we trained long short-term memory (LSTM recurrent neural networks with the training data provided by the DARPA / KDD Cup ’99 challenge. To identify suitable LSTM-RNN network parameters and structure we experimented with various network topologies. We found networks with four memory blocks containing two cells each offer a good compromise between computational cost and detection performance. We applied forget gates and shortcut connections respectively. A learning rate of 0.1 and up to 1,000 epochs showed good results. We tested the performance on all features and on extracted minimal feature sets respectively. We evaluated different feature sets for the detection of all attacks within one network and also to train networks specialised on individual attack classes. Our results show that the LSTM classifier provides superior performance in comparison to results previously published results of strong static classifiers. With 93.82% accuracy and 22.13 cost, LSTM outperforms the winning entries of the KDD Cup ’99 challenge by far. This is due to the fact that LSTM learns to look back in time and correlate consecutive connection records. For the first time ever, we have demonstrated the usefulness of LSTM networks to intrusion detection.
Intelligent Noise Removal from EMG Signal Using Focused Time-Lagged Recurrent Neural Network
Directory of Open Access Journals (Sweden)
S. N. Kale
2009-01-01
Full Text Available Electromyography (EMG signals can be used for clinical/biomedical application and modern human computer interaction. EMG signals acquire noise while traveling through tissue, inherent noise in electronics equipment, ambient noise, and so forth. ANN approach is studied for reduction of noise in EMG signal. In this paper, it is shown that Focused Time-Lagged Recurrent Neural Network (FTLRNN can elegantly solve to reduce the noise from EMG signal. After rigorous computer simulations, authors developed an optimal FTLRNN model, which removes the noise from the EMG signal. Results show that the proposed optimal FTLRNN model has an MSE (Mean Square Error as low as 0.000067 and 0.000048, correlation coefficient as high as 0.99950 and 0.99939 for noise signal and EMG signal, respectively, when validated on the test dataset. It is also noticed that the output of the estimated FTLRNN model closely follows the real one. This network is indeed robust as EMG signal tolerates the noise variance from 0.1 to 0.4 for uniform noise and 0.30 for Gaussian noise. It is clear that the training of the network is independent of specific partitioning of dataset. It is seen that the performance of the proposed FTLRNN model clearly outperforms the best Multilayer perceptron (MLP and Radial Basis Function NN (RBF models. The simple NN model such as the FTLRNN with single-hidden layer can be employed to remove noise from EMG signal.
Interactive natural language acquisition in a multi-modal recurrent neural architecture
Heinrich, Stefan; Wermter, Stefan
2018-01-01
For the complex human brain that enables us to communicate in natural language, we gathered good understandings of principles underlying language acquisition and processing, knowledge about sociocultural conditions, and insights into activity patterns in the brain. However, we were not yet able to understand the behavioural and mechanistic characteristics for natural language and how mechanisms in the brain allow to acquire and process language. In bridging the insights from behavioural psychology and neuroscience, the goal of this paper is to contribute a computational understanding of appropriate characteristics that favour language acquisition. Accordingly, we provide concepts and refinements in cognitive modelling regarding principles and mechanisms in the brain and propose a neurocognitively plausible model for embodied language acquisition from real-world interaction of a humanoid robot with its environment. In particular, the architecture consists of a continuous time recurrent neural network, where parts have different leakage characteristics and thus operate on multiple timescales for every modality and the association of the higher level nodes of all modalities into cell assemblies. The model is capable of learning language production grounded in both, temporal dynamic somatosensation and vision, and features hierarchical concept abstraction, concept decomposition, multi-modal integration, and self-organisation of latent representations.
Construction of Gene Regulatory Networks Using Recurrent Neural Networks and Swarm Intelligence.
Khan, Abhinandan; Mandal, Sudip; Pal, Rajat Kumar; Saha, Goutam
2016-01-01
We have proposed a methodology for the reverse engineering of biologically plausible gene regulatory networks from temporal genetic expression data. We have used established information and the fundamental mathematical theory for this purpose. We have employed the Recurrent Neural Network formalism to extract the underlying dynamics present in the time series expression data accurately. We have introduced a new hybrid swarm intelligence framework for the accurate training of the model parameters. The proposed methodology has been first applied to a small artificial network, and the results obtained suggest that it can produce the best results available in the contemporary literature, to the best of our knowledge. Subsequently, we have implemented our proposed framework on experimental (in vivo) datasets. Finally, we have investigated two medium sized genetic networks (in silico) extracted from GeneNetWeaver, to understand how the proposed algorithm scales up with network size. Additionally, we have implemented our proposed algorithm with half the number of time points. The results indicate that a reduction of 50% in the number of time points does not have an effect on the accuracy of the proposed methodology significantly, with a maximum of just over 15% deterioration in the worst case.
Using LSTM recurrent neural networks for monitoring the LHC superconducting magnets
Wielgosz, Maciej; Skoczeń, Andrzej; Mertik, Matej
2017-09-01
The superconducting LHC magnets are coupled with an electronic monitoring system which records and analyzes voltage time series reflecting their performance. A currently used system is based on a range of preprogrammed triggers which launches protection procedures when a misbehavior of the magnets is detected. All the procedures used in the protection equipment were designed and implemented according to known working scenarios of the system and are updated and monitored by human operators. This paper proposes a novel approach to monitoring and fault protection of the Large Hadron Collider (LHC) superconducting magnets which employs state-of-the-art Deep Learning algorithms. Consequently, the authors of the paper decided to examine the performance of LSTM recurrent neural networks for modeling of voltage time series of the magnets. In order to address this challenging task different network architectures and hyper-parameters were used to achieve the best possible performance of the solution. The regression results were measured in terms of RMSE for different number of future steps and history length taken into account for the prediction. The best result of RMSE = 0 . 00104 was obtained for a network of 128 LSTM cells within the internal layer and 16 steps history buffer.
Ni, Zhaoheng; Yuksel, Ahmet Cem; Ni, Xiuyan; Mandel, Michael I; Xie, Lei
2017-08-01
Brain fog, also known as confusion, is one of the main reasons for low performance in the learning process or any kind of daily task that involves and requires thinking. Detecting confusion in a human's mind in real time is a challenging and important task that can be applied to online education, driver fatigue detection and so on. In this paper, we apply Bidirectional LSTM Recurrent Neural Networks to classify students' confusion in watching online course videos from EEG data. The results show that Bidirectional LSTM model achieves the state-of-the-art performance compared with other machine learning approaches, and shows strong robustness as evaluated by cross-validation. We can predict whether or not a student is confused in the accuracy of 73.3%. Furthermore, we find the most important feature to detecting the brain confusion is the gamma 1 wave of EEG signal. Our results suggest that machine learning is a potentially powerful tool to model and understand brain activity.
Application of Recurrent Neural Networks on El Nino Impact on California Climate
Le, J.; El-Askary, H. M.; Allai, M.
2017-12-01
Following our successful paper on the application for the El Nino season of 2015-2016 over Southern California, we use recurrent neural networks (RNNs) to investigate the complex interactions between the long-term trend in dryness and a projected, short but intense, period of wetness due to the 2015-2016 El Niño. Although it was forecasted that this El Niño season would bring significant rainfall to the region, our long-term projections of the Palmer Z Index (PZI) showed a continuing drought trend. We achieved a statistically significant correlation of 0.610 between forecasted and observed PZI on the validation set for a lead time of 1 month. This gives strong confidence to the forecasted precipitation indicator. These predictions were bourne out in the resulting data. This paper details the expansion of our system to the climate of the entire California climate as a whole, dealing with inter-relationships and spatial variations within the state.
Local community detection as pattern restoration by attractor dynamics of recurrent neural networks.
Okamoto, Hiroshi
2016-08-01
Densely connected parts in networks are referred to as "communities". Community structure is a hallmark of a variety of real-world networks. Individual communities in networks form functional modules of complex systems described by networks. Therefore, finding communities in networks is essential to approaching and understanding complex systems described by networks. In fact, network science has made a great deal of effort to develop effective and efficient methods for detecting communities in networks. Here we put forward a type of community detection, which has been little examined so far but will be practically useful. Suppose that we are given a set of source nodes that includes some (but not all) of "true" members of a particular community; suppose also that the set includes some nodes that are not the members of this community (i.e., "false" members of the community). We propose to detect the community from this "imperfect" and "inaccurate" set of source nodes using attractor dynamics of recurrent neural networks. Community detection by the proposed method can be viewed as restoration of the original pattern from a deteriorated pattern, which is analogous to cue-triggered recall of short-term memory in the brain. We demonstrate the effectiveness of the proposed method using synthetic networks and real social networks for which correct communities are known. Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.
Chherawala, Youssouf; Roy, Partha Pratim; Cheriet, Mohamed
2016-12-01
The performance of handwriting recognition systems is dependent on the features extracted from the word image. A large body of features exists in the literature, but no method has yet been proposed to identify the most promising of these, other than a straightforward comparison based on the recognition rate. In this paper, we propose a framework for feature set evaluation based on a collaborative setting. We use a weighted vote combination of recurrent neural network (RNN) classifiers, each trained with a particular feature set. This combination is modeled in a probabilistic framework as a mixture model and two methods for weight estimation are described. The main contribution of this paper is to quantify the importance of feature sets through the combination weights, which reflect their strength and complementarity. We chose the RNN classifier because of its state-of-the-art performance. Also, we provide the first feature set benchmark for this classifier. We evaluated several feature sets on the IFN/ENIT and RIMES databases of Arabic and Latin script, respectively. The resulting combination model is competitive with state-of-the-art systems.
Emergence of unstable itinerant orbits in a recurrent neural network model
International Nuclear Information System (INIS)
Suemitsu, Yoshikazu; Nara, Shigetoshi
2005-01-01
A recurrent neural network model with time delay is investigated by numerical methods. The model functions as both conventional associative memory and also enables us to embed a new kind of memory attractor that cannot be realized in models without time delay, for example chain-ring attractors. This is attributed to the fact that the time delay extends the available state space dimension. The difference between the basin structures of chain-ring attractors and of isolated cycle attractors is investigated with respect to the two attractor pattern sets, random memory patterns and designed memory patterns with intended structures. Compared to isolated attractors with random memory patterns, the basins of chain-ring attractors are reduced considerably. Computer experiments confirm that the basin volume of each embedded chain-ring attractor shrinks and the emergence of unstable itinerant orbits in the outer state space of the memory attractor basins is discovered. The instability of such itinerant orbits is investigated. Results show that a 1-bit difference in initial conditions does not exceed 10% of a total dimension within 100 updating steps
A Recurrent Neural Network Approach to Rear Vehicle Detection Which Considered State Dependency
Directory of Open Access Journals (Sweden)
Kayichirou Inagaki
2003-08-01
Full Text Available Experimental vision-based detection often fails in cases when the acquired image quality is reduced by changing optical environments. In addition, the shape of vehicles in images that are taken from vision sensors change due to approaches by vehicle. Vehicle detection methods are required to perform successfully under these conditions. However, the conventional methods do not consider especially in rapidly varying by brightness conditions. We suggest a new detection method that compensates for those conditions in monocular vision-based vehicle detection. The suggested method employs a Recurrent Neural Network (RNN, which has been applied for spatiotemporal processing. The RNN is able to respond to consecutive scenes involving the target vehicle and can track the movements of the target by the effect of the past network states. The suggested method has a particularly beneficial effect in environments with sudden, extreme variations such as bright sunlight and shield. Finally, we demonstrate effectiveness by state-dependent of the RNN-based method by comparing its detection results with those of a Multi Layered Perceptron (MLP.
International Nuclear Information System (INIS)
Song Qiankun
2008-01-01
In this paper, the global exponential periodicity and stability of recurrent neural networks with time-varying delays are investigated by applying the idea of vector Lyapunov function, M-matrix theory and inequality technique. We assume neither the global Lipschitz conditions on these activation functions nor the differentiability on these time-varying delays, which were needed in other papers. Several novel criteria are found to ascertain the existence, uniqueness and global exponential stability of periodic solution for recurrent neural network with time-varying delays. Moreover, the exponential convergence rate index is estimated, which depends on the system parameters. Some previous results are improved and generalized, and an example is given to show the effectiveness of our method
Mandal, Sudip; Khan, Abhinandan; Saha, Goutam; Pal, Rajat K
2016-01-01
The accurate prediction of genetic networks using computational tools is one of the greatest challenges in the postgenomic era. Recurrent Neural Network is one of the most popular but simple approaches to model the network dynamics from time-series microarray data. To date, it has been successfully applied to computationally derive small-scale artificial and real-world genetic networks with high accuracy. However, they underperformed for large-scale genetic networks. Here, a new methodology has been proposed where a hybrid Cuckoo Search-Flower Pollination Algorithm has been implemented with Recurrent Neural Network. Cuckoo Search is used to search the best combination of regulators. Moreover, Flower Pollination Algorithm is applied to optimize the model parameters of the Recurrent Neural Network formalism. Initially, the proposed method is tested on a benchmark large-scale artificial network for both noiseless and noisy data. The results obtained show that the proposed methodology is capable of increasing the inference of correct regulations and decreasing false regulations to a high degree. Secondly, the proposed methodology has been validated against the real-world dataset of the DNA SOS repair network of Escherichia coli. However, the proposed method sacrifices computational time complexity in both cases due to the hybrid optimization process.
Directory of Open Access Journals (Sweden)
Bahita Mohamed
2011-01-01
Full Text Available In this work, we introduce an adaptive neural network controller for a class of nonlinear systems. The approach uses two Radial Basis Functions, RBF networks. The first RBF network is used to approximate the ideal control law which cannot be implemented since the dynamics of the system are unknown. The second RBF network is used for on-line estimating the control gain which is a nonlinear and unknown function of the states. The updating laws for the combined estimator and controller are derived through Lyapunov analysis. Asymptotic stability is established with the tracking errors converging to a neighborhood of the origin. Finally, the proposed method is applied to control and stabilize the inverted pendulum system.
Panda, Priyadarshini; Roy, Kaushik
2017-01-01
Synaptic Plasticity, the foundation for learning and memory formation in the human brain, manifests in various forms. Here, we combine the standard spike timing correlation based Hebbian plasticity with a non-Hebbian synaptic decay mechanism for training a recurrent spiking neural model to generate sequences. We show that inclusion of the adaptive decay of synaptic weights with standard STDP helps learn stable contextual dependencies between temporal sequences, while reducing the strong attractor states that emerge in recurrent models due to feedback loops. Furthermore, we show that the combined learning scheme suppresses the chaotic activity in the recurrent model substantially, thereby enhancing its' ability to generate sequences consistently even in the presence of perturbations.
Keshavarzi, Mahmoud; Goehring, Tobias; Zakis, Justin; Turner, Richard E.; Moore, Brian C. J.
2018-01-01
Despite great advances in hearing-aid technology, users still experience problems with noise in windy environments. The potential benefits of using a deep recurrent neural network (RNN) for reducing wind noise were assessed. The RNN was trained using recordings of the output of the two microphones of a behind-the-ear hearing aid in response to male and female speech at various azimuths in the presence of noise produced by wind from various azimuths with a velocity of 3 m/s, using the “clean” speech as a reference. A paired-comparison procedure was used to compare all possible combinations of three conditions for subjective intelligibility and for sound quality or comfort. The conditions were unprocessed noisy speech, noisy speech processed using the RNN, and noisy speech that was high-pass filtered (which also reduced wind noise). Eighteen native English-speaking participants were tested, nine with normal hearing and nine with mild-to-moderate hearing impairment. Frequency-dependent linear amplification was provided for the latter. Processing using the RNN was significantly preferred over no processing by both subject groups for both subjective intelligibility and sound quality, although the magnitude of the preferences was small. High-pass filtering (HPF) was not significantly preferred over no processing. Although RNN was significantly preferred over HPF only for sound quality for the hearing-impaired participants, for the results as a whole, there was a preference for RNN over HPF. Overall, the results suggest that reduction of wind noise using an RNN is possible and might have beneficial effects when used in hearing aids. PMID:29708061
Keshavarzi, Mahmoud; Goehring, Tobias; Zakis, Justin; Turner, Richard E; Moore, Brian C J
2018-01-01
Despite great advances in hearing-aid technology, users still experience problems with noise in windy environments. The potential benefits of using a deep recurrent neural network (RNN) for reducing wind noise were assessed. The RNN was trained using recordings of the output of the two microphones of a behind-the-ear hearing aid in response to male and female speech at various azimuths in the presence of noise produced by wind from various azimuths with a velocity of 3 m/s, using the "clean" speech as a reference. A paired-comparison procedure was used to compare all possible combinations of three conditions for subjective intelligibility and for sound quality or comfort. The conditions were unprocessed noisy speech, noisy speech processed using the RNN, and noisy speech that was high-pass filtered (which also reduced wind noise). Eighteen native English-speaking participants were tested, nine with normal hearing and nine with mild-to-moderate hearing impairment. Frequency-dependent linear amplification was provided for the latter. Processing using the RNN was significantly preferred over no processing by both subject groups for both subjective intelligibility and sound quality, although the magnitude of the preferences was small. High-pass filtering (HPF) was not significantly preferred over no processing. Although RNN was significantly preferred over HPF only for sound quality for the hearing-impaired participants, for the results as a whole, there was a preference for RNN over HPF. Overall, the results suggest that reduction of wind noise using an RNN is possible and might have beneficial effects when used in hearing aids.
Zhou, Caigen; Zeng, Xiaoqin; Luo, Chaomin; Zhang, Huaguang
In this paper, local bipolar auto-associative memories are presented based on discrete recurrent neural networks with a class of gain type activation function. The weight parameters of neural networks are acquired by a set of inequalities without the learning procedure. The global exponential stability criteria are established to ensure the accuracy of the restored patterns by considering time delays and external inputs. The proposed methodology is capable of effectively overcoming spurious memory patterns and achieving memory capacity. The effectiveness, robustness, and fault-tolerant capability are validated by simulated experiments.In this paper, local bipolar auto-associative memories are presented based on discrete recurrent neural networks with a class of gain type activation function. The weight parameters of neural networks are acquired by a set of inequalities without the learning procedure. The global exponential stability criteria are established to ensure the accuracy of the restored patterns by considering time delays and external inputs. The proposed methodology is capable of effectively overcoming spurious memory patterns and achieving memory capacity. The effectiveness, robustness, and fault-tolerant capability are validated by simulated experiments.
Linear headache: a recurrent unilateral head pain circumscribed in a line-shaped area.
Wang, Yu; Tian, Miao-Miao; Wang, Xian-Hong; Zhu, Xiao-Qun; Liu, Ying; Lu, Ya-Nan; Pan, Qing-Qing
2014-06-26
A headache circumscribed in a line-shaped area but not confined to the territory of one particular nerve had ever been described in Epicrania Fugax (EF) of which the head pain is moving and ultrashort. In a 25-month period from Feb 2012 to Mar 2014, we encountered 12 patients with a paroxysmal motionless head pain restricted in a linear trajectory. The head pain trajectory was similar to that of EF, but its all other features obviously different from those of EF. We named this distinctive but undescribed type of headache linear headache (LH). A detailed clinical feature of the headache was obtained in all cases to differentiate with EF, trigeminal autonomic cephalalgias (TACs) and cranial neuralgia. Similarities and differences in clinical features were compared between LH and migraine. The twelve LH patients (mean age 43.9 ± 12.2) complained of a recurrent, moderate to severe, distending (n = 9), pressure-like (n = 3) or pulsating (n = 3) pain within a strictly unilateral line-shaped area. The painful line is distributed from occipital or occipitocervical region to the ipsilateral eye (n = 5), forehead (n = 6) or parietal region (n = 1). The pain line has a trajecory similar to that of EF but no characteristics of moving. The headache duration would be ranged from five minutes to three days, but usually from half day to one day in most cases (n = 8). Six patients had the accompaniment of nausea with or without vomiting, and two patients had the accompaniment of ipsilateral dizziness. The attacks could be either spontaneous (n = 10) or triggered by noise, depression and resting after physical activity (n = 1), or by stress and staying up late (n = 1). The frequency of attacks was variable. The patients had well response to flunarizine, sodium valproate and amitriptyline but not to carbamazepine or oxcarbazepine. LH is different from EF, trigeminal autonomic cephalalgias (TACs) and cranial neuralgia, but it had couple of features similar to that of migraine. The
Non-Linear State Estimation Using Pre-Trained Neural Networks
DEFF Research Database (Denmark)
Bayramoglu, Enis; Andersen, Nils Axel; Ravn, Ole
2010-01-01
effecting the transformation. This function is approximated by a neural network using offline training. The training is based on monte carlo sampling. A way to obtain parametric distributions of flexible shape to be used easily with these networks is also presented. The method can also be used to improve...... other parametric methods around regions with strong non-linearities by including them inside the network....
Bengoetxea, Ana; Leurs, Françoise; Hoellinger, Thomas; Cebolla, Ana M; Dan, Bernard; McIntyre, Joseph; Cheron, Guy
2014-01-01
In this study we employed a dynamic recurrent neural network (DRNN) in a novel fashion to reveal characteristics of control modules underlying the generation of muscle activations when drawing figures with the outstretched arm. We asked healthy human subjects to perform four different figure-eight movements in each of two workspaces (frontal plane and sagittal plane). We then trained a DRNN to predict the movement of the wrist from information in the EMG signals from seven different muscles. We trained different instances of the same network on a single movement direction, on all four movement directions in a single movement plane, or on all eight possible movement patterns and looked at the ability of the DRNN to generalize and predict movements for trials that were not included in the training set. Within a single movement plane, a DRNN trained on one movement direction was not able to predict movements of the hand for trials in the other three directions, but a DRNN trained simultaneously on all four movement directions could generalize across movement directions within the same plane. Similarly, the DRNN was able to reproduce the kinematics of the hand for both movement planes, but only if it was trained on examples performed in each one. As we will discuss, these results indicate that there are important dynamical constraints on the mapping of EMG to hand movement that depend on both the time sequence of the movement and on the anatomical constraints of the musculoskeletal system. In a second step, we injected EMG signals constructed from different synergies derived by the PCA in order to identify the mechanical significance of each of these components. From these results, one can surmise that discrete-rhythmic movements may be constructed from three different fundamental modules, one regulating the co-activation of all muscles over the time span of the movement and two others elliciting patterns of reciprocal activation operating in orthogonal directions.
Jauregi Unanue, Iñigo; Zare Borzeshi, Ehsan; Piccardi, Massimo
2017-12-01
Previous state-of-the-art systems on Drug Name Recognition (DNR) and Clinical Concept Extraction (CCE) have focused on a combination of text "feature engineering" and conventional machine learning algorithms such as conditional random fields and support vector machines. However, developing good features is inherently heavily time-consuming. Conversely, more modern machine learning approaches such as recurrent neural networks (RNNs) have proved capable of automatically learning effective features from either random assignments or automated word "embeddings". (i) To create a highly accurate DNR and CCE system that avoids conventional, time-consuming feature engineering. (ii) To create richer, more specialized word embeddings by using health domain datasets such as MIMIC-III. (iii) To evaluate our systems over three contemporary datasets. Two deep learning methods, namely the Bidirectional LSTM and the Bidirectional LSTM-CRF, are evaluated. A CRF model is set as the baseline to compare the deep learning systems to a traditional machine learning approach. The same features are used for all the models. We have obtained the best results with the Bidirectional LSTM-CRF model, which has outperformed all previously proposed systems. The specialized embeddings have helped to cover unusual words in DrugBank and MedLine, but not in the i2b2/VA dataset. We present a state-of-the-art system for DNR and CCE. Automated word embeddings has allowed us to avoid costly feature engineering and achieve higher accuracy. Nevertheless, the embeddings need to be retrained over datasets that are adequate for the domain, in order to adequately cover the domain-specific vocabulary. Copyright © 2017 Elsevier Inc. All rights reserved.
Learning a Transferable Change Rule from a Recurrent Neural Network for Land Cover Change Detection
Directory of Open Access Journals (Sweden)
Haobo Lyu
2016-06-01
Full Text Available When exploited in remote sensing analysis, a reliable change rule with transfer ability can detect changes accurately and be applied widely. However, in practice, the complexity of land cover changes makes it difficult to use only one change rule or change feature learned from a given multi-temporal dataset to detect any other new target images without applying other learning processes. In this study, we consider the design of an efficient change rule having transferability to detect both binary and multi-class changes. The proposed method relies on an improved Long Short-Term Memory (LSTM model to acquire and record the change information of long-term sequence remote sensing data. In particular, a core memory cell is utilized to learn the change rule from the information concerning binary changes or multi-class changes. Three gates are utilized to control the input, output and update of the LSTM model for optimization. In addition, the learned rule can be applied to detect changes and transfer the change rule from one learned image to another new target multi-temporal image. In this study, binary experiments, transfer experiments and multi-class change experiments are exploited to demonstrate the superiority of our method. Three contributions of this work can be summarized as follows: (1 the proposed method can learn an effective change rule to provide reliable change information for multi-temporal images; (2 the learned change rule has good transferability for detecting changes in new target images without any extra learning process, and the new target images should have a multi-spectral distribution similar to that of the training images; and (3 to the authors’ best knowledge, this is the first time that deep learning in recurrent neural networks is exploited for change detection. In addition, under the framework of the proposed method, changes can be detected under both binary detection and multi-class change detection.
Cocos, Anne; Fiks, Alexander G; Masino, Aaron J
2017-07-01
Social media is an important pharmacovigilance data source for adverse drug reaction (ADR) identification. Human review of social media data is infeasible due to data quantity, thus natural language processing techniques are necessary. Social media includes informal vocabulary and irregular grammar, which challenge natural language processing methods. Our objective is to develop a scalable, deep-learning approach that exceeds state-of-the-art ADR detection performance in social media. We developed a recurrent neural network (RNN) model that labels words in an input sequence with ADR membership tags. The only input features are word-embedding vectors, which can be formed through task-independent pretraining or during ADR detection training. Our best-performing RNN model used pretrained word embeddings created from a large, non-domain-specific Twitter dataset. It achieved an approximate match F-measure of 0.755 for ADR identification on the dataset, compared to 0.631 for a baseline lexicon system and 0.65 for the state-of-the-art conditional random field model. Feature analysis indicated that semantic information in pretrained word embeddings boosted sensitivity and, combined with contextual awareness captured in the RNN, precision. Our model required no task-specific feature engineering, suggesting generalizability to additional sequence-labeling tasks. Learning curve analysis showed that our model reached optimal performance with fewer training examples than the other models. ADR detection performance in social media is significantly improved by using a contextually aware model and word embeddings formed from large, unlabeled datasets. The approach reduces manual data-labeling requirements and is scalable to large social media datasets. © The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Directory of Open Access Journals (Sweden)
Tatsuro Yamada
2016-07-01
Full Text Available To work cooperatively with humans by using language, robots must not only acquire a mapping between language and their behavior but also autonomously utilize the mapping in appropriate contexts of interactive tasks online. To this end, we propose a novel learning method linking language to robot behavior by means of a recurrent neural network. In this method, the network learns from correct examples of the imposed task that are given not as explicitly separated sets of language and behavior but as sequential data constructed from the actual temporal flow of the task. By doing this, the internal dynamics of the network models both language--behavior relationships and the temporal patterns of interaction. Here, ``internal dynamics'' refers to the time development of the system defined on the fixed-dimensional space of the internal states of the context layer. Thus, in the execution phase, by constantly representing where in the interaction context it is as its current state, the network autonomously switches between recognition and generation phases without any explicit signs and utilizes the acquired mapping in appropriate contexts. To evaluate our method, we conducted an experiment in which a robot generates appropriate behavior responding to a human's linguistic instruction. After learning, the network actually formed the attractor structure representing both language--behavior relationships and the task's temporal pattern in its internal dynamics. In the dynamics, language--behavior mapping was achieved by the branching structure. Repetition of human's instruction and robot's behavioral response was represented as the cyclic structure, and besides, waiting to a subsequent instruction was represented as the fixed-point attractor. Thanks to this structure, the robot was able to interact online with a human concerning the given task by autonomously switching phases.
Yamada, Tatsuro; Murata, Shingo; Arie, Hiroaki; Ogata, Tetsuya
2016-01-01
To work cooperatively with humans by using language, robots must not only acquire a mapping between language and their behavior but also autonomously utilize the mapping in appropriate contexts of interactive tasks online. To this end, we propose a novel learning method linking language to robot behavior by means of a recurrent neural network. In this method, the network learns from correct examples of the imposed task that are given not as explicitly separated sets of language and behavior but as sequential data constructed from the actual temporal flow of the task. By doing this, the internal dynamics of the network models both language-behavior relationships and the temporal patterns of interaction. Here, "internal dynamics" refers to the time development of the system defined on the fixed-dimensional space of the internal states of the context layer. Thus, in the execution phase, by constantly representing where in the interaction context it is as its current state, the network autonomously switches between recognition and generation phases without any explicit signs and utilizes the acquired mapping in appropriate contexts. To evaluate our method, we conducted an experiment in which a robot generates appropriate behavior responding to a human's linguistic instruction. After learning, the network actually formed the attractor structure representing both language-behavior relationships and the task's temporal pattern in its internal dynamics. In the dynamics, language-behavior mapping was achieved by the branching structure. Repetition of human's instruction and robot's behavioral response was represented as the cyclic structure, and besides, waiting to a subsequent instruction was represented as the fixed-point attractor. Thanks to this structure, the robot was able to interact online with a human concerning the given task by autonomously switching phases.
De-identification of clinical notes via recurrent neural network and conditional random field.
Liu, Zengjian; Tang, Buzhou; Wang, Xiaolong; Chen, Qingcai
2017-11-01
De-identification, identifying information from data, such as protected health information (PHI) present in clinical data, is a critical step to enable data to be shared or published. The 2016 Centers of Excellence in Genomic Science (CEGS) Neuropsychiatric Genome-scale and RDOC Individualized Domains (N-GRID) clinical natural language processing (NLP) challenge contains a de-identification track in de-identifying electronic medical records (EMRs) (i.e., track 1). The challenge organizers provide 1000 annotated mental health records for this track, 600 out of which are used as a training set and 400 as a test set. We develop a hybrid system for the de-identification task on the training set. Firstly, four individual subsystems, that is, a subsystem based on bidirectional LSTM (long-short term memory, a variant of recurrent neural network), a subsystem-based on bidirectional LSTM with features, a subsystem based on conditional random field (CRF) and a rule-based subsystem, are used to identify PHI instances. Then, an ensemble learning-based classifiers is deployed to combine all PHI instances predicted by above three machine learning-based subsystems. Finally, the results of the ensemble learning-based classifier and the rule-based subsystem are merged together. Experiments conducted on the official test set show that our system achieves the highest micro F1-scores of 93.07%, 91.43% and 95.23% under the "token", "strict" and "binary token" criteria respectively, ranking first in the 2016 CEGS N-GRID NLP challenge. In addition, on the dataset of 2014 i2b2 NLP challenge, our system achieves the highest micro F1-scores of 96.98%, 95.11% and 98.28% under the "token", "strict" and "binary token" criteria respectively, outperforming other state-of-the-art systems. All these experiments prove the effectiveness of our proposed method. Copyright © 2017. Published by Elsevier Inc.
Lin, Chin-Teng; Wu, Rui-Cheng; Chang, Jyh-Yeong; Liang, Sheng-Fu
2004-02-01
In this paper, a new technique for the Chinese text-to-speech (TTS) system is proposed. Our major effort focuses on the prosodic information generation. New methodologies for constructing fuzzy rules in a prosodic model simulating human's pronouncing rules are developed. The proposed Recurrent Fuzzy Neural Network (RFNN) is a multilayer recurrent neural network (RNN) which integrates a Self-cOnstructing Neural Fuzzy Inference Network (SONFIN) into a recurrent connectionist structure. The RFNN can be functionally divided into two parts. The first part adopts the SONFIN as a prosodic model to explore the relationship between high-level linguistic features and prosodic information based on fuzzy inference rules. As compared to conventional neural networks, the SONFIN can always construct itself with an economic network size in high learning speed. The second part employs a five-layer network to generate all prosodic parameters by directly using the prosodic fuzzy rules inferred from the first part as well as other important features of syllables. The TTS system combined with the proposed method can behave not only sandhi rules but also the other prosodic phenomena existing in the traditional TTS systems. Moreover, the proposed scheme can even find out some new rules about prosodic phrase structure. The performance of the proposed RFNN-based prosodic model is verified by imbedding it into a Chinese TTS system with a Chinese monosyllable database based on the time-domain pitch synchronous overlap add (TD-PSOLA) method. Our experimental results show that the proposed RFNN can generate proper prosodic parameters including pitch means, pitch shapes, maximum energy levels, syllable duration, and pause duration. Some synthetic sounds are online available for demonstration.
Directory of Open Access Journals (Sweden)
Sakyasingha eDasgupta
2015-09-01
Full Text Available Walking animals, like stick insects, cockroaches or ants, demonstrate a fascinating range of locomotive abilities and complex behaviors. The locomotive behaviors can consist of a variety of walking patterns along with adaptation that allow the animals to deal with changes in environmental conditions, like uneven terrains, gaps, obstacles etc. Biological study has revealed that such complex behaviors are a result of a combination of biomechanics and neural mechanism thus representing the true nature of embodied interactions. While the biomechanics helps maintain flexibility and sustain a variety of movements, the neural mechanisms generate movements while making appropriate predictions crucial for achieving adaptation. Such predictions or planning ahead can be achieved by way of internal models that are grounded in the overall behavior of the animal. Inspired by these findings, we present here, an artificial bio-inspired walking system which effectively combines biomechanics (in terms of the body and leg structures with the underlying neural mechanisms. The neural mechanisms consist of 1 central pattern generator based control for generating basic rhythmic patterns and coordinated movements, 2 distributed (at each leg recurrent neural network based adaptive forward models with efference copies as internal models for sensory predictions and instantaneous state estimations, and 3 searching and elevation control for adapting the movement of an individual leg to deal with different environmental conditions. Using simulations we show that this bio-inspired approach with adaptive internal models allows the walking robot to perform complex locomotive behaviors as observed in insects, including walking on undulated terrains, crossing large gaps as well as climbing over high obstacles. Furthermore we demonstrate that the newly developed recurrent network based approach to sensorimotor prediction outperforms the previous state of the art adaptive neuron
Directory of Open Access Journals (Sweden)
S. N. Naikwad
2009-01-01
Full Text Available A focused time lagged recurrent neural network (FTLR NN with gamma memory filter is designed to learn the subtle complex dynamics of a typical CSTR process. Continuous stirred tank reactor exhibits complex nonlinear operations where reaction is exothermic. It is noticed from literature review that process control of CSTR using neuro-fuzzy systems was attempted by many, but optimal neural network model for identification of CSTR process is not yet available. As CSTR process includes temporal relationship in the input-output mappings, time lagged recurrent neural network is particularly used for identification purpose. The standard back propagation algorithm with momentum term has been proposed in this model. The various parameters like number of processing elements, number of hidden layers, training and testing percentage, learning rule and transfer function in hidden and output layer are investigated on the basis of performance measures like MSE, NMSE, and correlation coefficient on testing data set. Finally effects of different norms are tested along with variation in gamma memory filter. It is demonstrated that dynamic NN model has a remarkable system identification capability for the problems considered in this paper. Thus FTLR NN with gamma memory filter can be used to learn underlying highly nonlinear dynamics of the system, which is a major contribution of this paper.
Faulty node detection in wireless sensor networks using a recurrent neural network
Atiga, Jamila; Mbarki, Nour Elhouda; Ejbali, Ridha; Zaied, Mourad
2018-04-01
The wireless sensor networks (WSN) consist of a set of sensors that are more and more used in surveillance applications on a large scale in different areas: military, Environment, Health ... etc. Despite the minimization and the reduction of the manufacturing costs of the sensors, they can operate in places difficult to access without the possibility of reloading of battery, they generally have limited resources in terms of power of emission, of processing capacity, data storage and energy. These sensors can be used in a hostile environment, such as, for example, on a field of battle, in the presence of fires, floods, earthquakes. In these environments the sensors can fail, even in a normal operation. It is therefore necessary to develop algorithms tolerant and detection of defects of the nodes for the network of sensor without wires, therefore, the faults of the sensor can reduce the quality of the surveillance if they are not detected. The values that are measured by the sensors are used to estimate the state of the monitored area. We used the Non-linear Auto- Regressive with eXogeneous (NARX), the recursive architecture of the neural network, to predict the state of a node of a sensor from the previous values described by the functions of time series. The experimental results have verified that the prediction of the State is enhanced by our proposed model.
International Nuclear Information System (INIS)
Balasubramaniam, P.; Kalpana, M.; Rakkiyappan, R.
2012-01-01
Fuzzy cellular neural networks (FCNNs) are special kinds of cellular neural networks (CNNs). Each cell in an FCNN contains fuzzy operating abilities. The entire network is governed by cellular computing laws. The design of FCNNs is based on fuzzy local rules. In this paper, a linear matrix inequality (LMI) approach for synchronization control of FCNNs with mixed delays is investigated. Mixed delays include discrete time-varying delays and unbounded distributed delays. A dynamic control scheme is proposed to achieve the synchronization between a drive network and a response network. By constructing the Lyapunov—Krasovskii functional which contains a triple-integral term and the free-weighting matrices method an improved delay-dependent stability criterion is derived in terms of LMIs. The controller can be easily obtained by solving the derived LMIs. A numerical example and its simulations are presented to illustrate the effectiveness of the proposed method. (interdisciplinary physics and related areas of science and technology)
Multiple Linear Regression Model Based on Neural Network and Its Application in the MBR Simulation
Directory of Open Access Journals (Sweden)
Chunqing Li
2012-01-01
Full Text Available The computer simulation of the membrane bioreactor MBR has become the research focus of the MBR simulation. In order to compensate for the defects, for example, long test period, high cost, invisible equipment seal, and so forth, on the basis of conducting in-depth study of the mathematical model of the MBR, combining with neural network theory, this paper proposed a three-dimensional simulation system for MBR wastewater treatment, with fast speed, high efficiency, and good visualization. The system is researched and developed with the hybrid programming of VC++ programming language and OpenGL, with a multifactor linear regression model of affecting MBR membrane fluxes based on neural network, applying modeling method of integer instead of float and quad tree recursion. The experiments show that the three-dimensional simulation system, using the above models and methods, has the inspiration and reference for the future research and application of the MBR simulation technology.
Ling, Hong; Samarasinghe, Sandhya; Kulasiri, Don
2013-12-01
Understanding the control of cellular networks consisting of gene and protein interactions and their emergent properties is a central activity of Systems Biology research. For this, continuous, discrete, hybrid, and stochastic methods have been proposed. Currently, the most common approach to modelling accurate temporal dynamics of networks is ordinary differential equations (ODE). However, critical limitations of ODE models are difficulty in kinetic parameter estimation and numerical solution of a large number of equations, making them more suited to smaller systems. In this article, we introduce a novel recurrent artificial neural network (RNN) that addresses above limitations and produces a continuous model that easily estimates parameters from data, can handle a large number of molecular interactions and quantifies temporal dynamics and emergent systems properties. This RNN is based on a system of ODEs representing molecular interactions in a signalling network. Each neuron represents concentration change of one molecule represented by an ODE. Weights of the RNN correspond to kinetic parameters in the system and can be adjusted incrementally during network training. The method is applied to the p53-Mdm2 oscillation system - a crucial component of the DNA damage response pathways activated by a damage signal. Simulation results indicate that the proposed RNN can successfully represent the behaviour of the p53-Mdm2 oscillation system and solve the parameter estimation problem with high accuracy. Furthermore, we presented a modified form of the RNN that estimates parameters and captures systems dynamics from sparse data collected over relatively large time steps. We also investigate the robustness of the p53-Mdm2 system using the trained RNN under various levels of parameter perturbation to gain a greater understanding of the control of the p53-Mdm2 system. Its outcomes on robustness are consistent with the current biological knowledge of this system. As more
Samarasinghe, S; Ling, H
In this paper, we show how to extend our previously proposed novel continuous time Recurrent Neural Networks (RNN) approach that retains the advantage of continuous dynamics offered by Ordinary Differential Equations (ODE) while enabling parameter estimation through adaptation, to larger signalling networks using a modular approach. Specifically, the signalling network is decomposed into several sub-models based on important temporal events in the network. Each sub-model is represented by the proposed RNN and trained using data generated from the corresponding ODE model. Trained sub-models are assembled into a whole system RNN which is then subjected to systems dynamics and sensitivity analyses. The concept is illustrated by application to G1/S transition in cell cycle using Iwamoto et al. (2008) ODE model. We decomposed the G1/S network into 3 sub-models: (i) E2F transcription factor release; (ii) E2F and CycE positive feedback loop for elevating cyclin levels; and (iii) E2F and CycA negative feedback to degrade E2F. The trained sub-models accurately represented system dynamics and parameters were in good agreement with the ODE model. The whole system RNN however revealed couple of parameters contributing to compounding errors due to feedback and required refinement to sub-model 2. These related to the reversible reaction between CycE/CDK2 and p27, its inhibitor. The revised whole system RNN model very accurately matched dynamics of the ODE system. Local sensitivity analysis of the whole system model further revealed the most dominant influence of the above two parameters in perturbing G1/S transition, giving support to a recent hypothesis that the release of inhibitor p27 from Cyc/CDK complex triggers cell cycle stage transition. To make the model useful in a practical setting, we modified each RNN sub-model with a time relay switch to facilitate larger interval input data (≈20min) (original model used data for 30s or less) and retrained them that produced
Mohammadzadeh, Ardashir; Ghaemi, Sehraneh
2015-09-01
This paper proposes a novel approach for training of proposed recurrent hierarchical interval type-2 fuzzy neural networks (RHT2FNN) based on the square-root cubature Kalman filters (SCKF). The SCKF algorithm is used to adjust the premise part of the type-2 FNN and the weights of defuzzification and the feedback weights. The recurrence property in the proposed network is the output feeding of each membership function to itself. The proposed RHT2FNN is employed in the sliding mode control scheme for the synchronization of chaotic systems. Unknown functions in the sliding mode control approach are estimated by RHT2FNN. Another application of the proposed RHT2FNN is the identification of dynamic nonlinear systems. The effectiveness of the proposed network and its learning algorithm is verified by several simulation examples. Furthermore, the universal approximation of RHT2FNNs is also shown. Copyright © 2015 ISA. Published by Elsevier Ltd. All rights reserved.
Lu, Weizhao; Huang, Chunhui; Hou, Kun; Shi, Liting; Zhao, Huihui; Li, Zhengmei; Qiu, Jianfeng
2018-05-01
In continuous-variable quantum key distribution (CV-QKD), weak signal carrying information transmits from Alice to Bob; during this process it is easily influenced by unknown noise which reduces signal-to-noise ratio, and strongly impacts reliability and stability of the communication. Recurrent quantum neural network (RQNN) is an artificial neural network model which can perform stochastic filtering without any prior knowledge of the signal and noise. In this paper, a modified RQNN algorithm with expectation maximization algorithm is proposed to process the signal in CV-QKD, which follows the basic rule of quantum mechanics. After RQNN, noise power decreases about 15 dBm, coherent signal recognition rate of RQNN is 96%, quantum bit error rate (QBER) drops to 4%, which is 6.9% lower than original QBER, and channel capacity is notably enlarged.
Matos, Sérgio; Antunes, Rui
2017-12-13
Curation of protein interactions from scientific articles is an important task, since interaction networks are essential for the understanding of biological processes associated with disease or pharmacological action for example. However, the increase in the number of publications that potentially contain relevant information turns this into a very challenging and expensive task. In this work we used a convolutional recurrent neural network for identifying relevant articles for extracting information regarding protein interactions. Using the BioCreative III Article Classification Task dataset, we achieved an area under the precision-recall curve of 0.715 and a Matthew's correlation coefficient of 0.600, which represents an improvement over previous works.
International Nuclear Information System (INIS)
Ali, M. Syed
2011-01-01
In this paper, the global stability of Takagi—Sugeno (TS) uncertain stochastic fuzzy recurrent neural networks with discrete and distributed time-varying delays (TSUSFRNNs) is considered. A novel LMI-based stability criterion is obtained by using Lyapunov functional theory to guarantee the asymptotic stability of TSUSFRNNs. The proposed stability conditions are demonstrated through numerical examples. Furthermore, the supplementary requirement that the time derivative of time-varying delays must be smaller than one is removed. Comparison results are demonstrated to show that the proposed method is more able to guarantee the widest stability region than the other methods available in the existing literature. (general)
International Nuclear Information System (INIS)
Hannen, Jennifer C; Buckner, Gregory D; Crews, John H
2012-01-01
This paper introduces an indirect intelligent sliding mode controller (IISMC) for shape memory alloy (SMA) actuators, specifically a flexible beam deflected by a single offset SMA tendon. The controller manipulates applied voltage, which alters SMA tendon temperature to track reference bending angles. A hysteretic recurrent neural network (HRNN) captures the nonlinear, hysteretic relationship between SMA temperature and bending angle. The variable structure control strategy provides robustness to model uncertainties and parameter variations, while effectively compensating for system nonlinearities, achieving superior tracking compared to an optimized PI controller. (paper)
Pre-Trained Neural Networks used for Non-Linear State Estimation
DEFF Research Database (Denmark)
Bayramoglu, Enis; Andersen, Nils Axel; Ravn, Ole
2011-01-01
of the paramters in the distribution. This transformation is approximated by a neural network using offline training, which is based on monte carlo sampling. In the paper, there will also be presented a method to construct a flexible distributions well suited for covering the effect of the non-linearities......The paper focuses on nonlinear state estimation assuming non-Gaussian distributions of the states and the disturbances. The posterior distribution and the aposteriori distribution is described by a chosen family of paramtric distributions. The state transformation then results in a transformation...
Neural Network Control for the Linear Motion of a Spherical Mobile Robot
Directory of Open Access Journals (Sweden)
Yao Cai
2011-09-01
Full Text Available This paper discussed the stabilization and position tracking control of the linear motion of an underactuated spherical robot. By considering the actuator dynamics, a complete dynamic model of the robot is deduced, which is a complex third order, two variables nonlinear differential system and those two variables have strong coupling due to the mechanical structure of the robot. Different from traditional treatments, no linearization is applied to this system but a single‐input multiple‐output PID (SIMO_PID controller is designed by adopting a six‐input single‐ output CMAC_GBF (Cerebellar Model Articulation Controller with General Basis Function neural network to compensate the actuator nonlinearity and the credit assignment (CA learning method to obtain faster convergence of CMAC_GBF. The proposed controller is generalizable to other single‐input multiple‐output system with good real‐time capability. Simulations in Matlab are used to validate the control effects.
Doubly Periodic Traveling Waves in a Cellular Neural Network with Linear Reaction
Directory of Open Access Journals (Sweden)
Lin JianJhong
2009-01-01
Full Text Available Szekeley observed that the dynamic pattern of the locomotion of salamanders can be explained by periodic vector sequences generated by logical neural networks. Such sequences can mathematically be described by "doubly periodic traveling waves" and therefore it is of interest to propose dynamic models that may produce such waves. One such dynamic network model is built here based on reaction-diffusion principles and a complete discussion is given for the existence of doubly periodic waves as outputs. Since there are 2 parameters in our model and 4 a priori unknown parameters involved in our search of solutions, our results are nontrivial. The reaction term in our model is a linear function and hence our results can also be interpreted as existence criteria for solutions of a nontrivial linear problem depending on 6 parameters.
Kabeshova, Anastasiia; Launay, Cyrille P; Gromov, Vasilii A; Annweiler, Cédric; Fantino, Bruno; Beauchet, Olivier
2015-04-01
Identification of the risk of recurrent falls is complex in older adults. The aim of this study was to examine the efficiency of 3 artificial neural networks (ANNs: multilayer perceptron [MLP], modified MLP, and neuroevolution of augmenting topologies [NEAT]) for the classification of recurrent fallers and nonrecurrent fallers using a set of clinical characteristics corresponding to risk factors of falls measured among community-dwelling older adults. Based on a cross-sectional design, 3289 community-dwelling volunteers aged 65 and older were recruited. Age, gender, body mass index (BMI), number of drugs daily taken, use of psychoactive drugs, diphosphonate, calcium, vitamin D supplements and walking aid, fear of falling, distance vision score, Timed Up and Go (TUG) score, lower-limb proprioception, handgrip strength, depressive symptoms, cognitive disorders, and history of falls were recorded. Participants were separated into 2 groups based on the number of falls that occurred over the past year: 0 or 1 fall and 2 or more falls. In addition, total population was separated into training and testing subgroups for ANN analysis. Among 3289 participants, 18.9% (n = 622) were recurrent fallers. NEAT, using 15 clinical characteristics (ie, use of walking aid, fear of falling, use of calcium, depression, use of vitamin D supplements, female, cognitive disorders, BMI 4, vision score 9 seconds, handgrip strength score ≤29 (N), and age ≥75 years), showed the best efficiency for identification of recurrent fallers, sensitivity (80.42%), specificity (92.54%), positive predictive value (84.38), negative predictive value (90.34), accuracy (88.39), and Cohen κ (0.74), compared with MLP and modified MLP. NEAT, using a set of 15 clinical characteristics, was an efficient ANN for the identification of recurrent fallers in older community-dwellers. Copyright © 2015 AMDA – The Society for Post-Acute and Long-Term Care Medicine. Published by Elsevier Inc. All rights reserved.
Directory of Open Access Journals (Sweden)
Claudimar Pereira da Veiga
2012-08-01
Full Text Available The importance of demand forecasting as a management tool is a well documented issue. However, it is difficult to measure costs generated by forecasting errors and to find a model that assimilate the detailed operation of each company adequately. In general, when linear models fail in the forecasting process, more complex nonlinear models are considered. Although some studies comparing traditional models and neural networks have been conducted in the literature, the conclusions are usually contradictory. In this sense, the objective was to compare the accuracy of linear methods and neural networks with the current method used by the company. The results of this analysis also served as input to evaluate influence of errors in demand forecasting on the financial performance of the company. The study was based on historical data from five groups of food products, from 2004 to 2008. In general, one can affirm that all models tested presented good results (much better than the current forecasting method used, with mean absolute percent error (MAPE around 10%. The total financial impact for the company was 6,05% on annual sales.
A Study of Recurrent and Convolutional Neural Networks in the Native Language Identification Task
Werfelmann, Robert
2018-01-01
around the world. The neural network models consisted of Long Short-Term Memory and Convolutional networks using the sentences of each document as the input. Additional statistical features were generated from the text to complement the predictions
A Maple package for computing Groebner bases for linear recurrence relations
International Nuclear Information System (INIS)
Gerdt, Vladimir P.; Robertz, Daniel
2006-01-01
A Maple package for computing Groebner bases of linear difference ideals is described. The underlying algorithm is based on Janet and Janet-like monomial divisions associated with finite difference operators. The package can be used, for example, for automatic generation of difference schemes for linear partial differential equations and for reduction of multiloop Feynman integrals. These two possible applications are illustrated by simple examples of the Laplace equation and a one-loop scalar integral of propagator type
A Maple package for computing Groebner bases for linear recurrence relations
Energy Technology Data Exchange (ETDEWEB)
Gerdt, Vladimir P. [Laboratory of Information Technologies, Joint Institute for Nuclear Research, 141980 Dubna (Russian Federation)]. E-mail: gerdt@jinr.ru; Robertz, Daniel [Lehrstuhl B fuer Mathematik, RWTH Aachen, Templergraben 64, D-52062 Aachen (Germany)]. E-mail: daniel@momo.math.rwth-aachen.de
2006-04-01
A Maple package for computing Groebner bases of linear difference ideals is described. The underlying algorithm is based on Janet and Janet-like monomial divisions associated with finite difference operators. The package can be used, for example, for automatic generation of difference schemes for linear partial differential equations and for reduction of multiloop Feynman integrals. These two possible applications are illustrated by simple examples of the Laplace equation and a one-loop scalar integral of propagator type.
International Nuclear Information System (INIS)
Zio, Enrico; Pedroni, Nicola; Broggi, Matteo; Golea, Lucia Roxana
2009-01-01
In this paper, an infinite impulse response locally recurrent neural network (IIR-LRNN) is employed for modelling the dynamics of the Lead Bismuth Eutectic eXperimental Accelerator Driven System (LBE-XADS). The network is trained by recursive back-propagation (RBP) and its ability in estimating transients is tested under various conditions. The results demonstrate the robustness of the locally recurrent scheme in the reconstruction of complex nonlinear dynamic relationships
Uca; Toriman, Ekhwan; Jaafar, Othman; Maru, Rosmini; Arfan, Amal; Saleh Ahmar, Ansari
2018-01-01
Prediction of suspended sediment discharge in a catchments area is very important because it can be used to evaluation the erosion hazard, management of its water resources, water quality, hydrology project management (dams, reservoirs, and irrigation) and to determine the extent of the damage that occurred in the catchments. Multiple Linear Regression analysis and artificial neural network can be used to predict the amount of daily suspended sediment discharge. Regression analysis using the least square method, whereas artificial neural networks using Radial Basis Function (RBF) and feedforward multilayer perceptron with three learning algorithms namely Levenberg-Marquardt (LM), Scaled Conjugate Descent (SCD) and Broyden-Fletcher-Goldfarb-Shanno Quasi-Newton (BFGS). The number neuron of hidden layer is three to sixteen, while in output layer only one neuron because only one output target. The mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R2 ) and coefficient of efficiency (CE) of the multiple linear regression (MLRg) value Model 2 (6 input variable independent) has the lowest the value of MAE and RMSE (0.0000002 and 13.6039) and highest R2 and CE (0.9971 and 0.9971). When compared between LM, SCG and RBF, the BFGS model structure 3-7-1 is the better and more accurate to prediction suspended sediment discharge in Jenderam catchment. The performance value in testing process, MAE and RMSE (13.5769 and 17.9011) is smallest, meanwhile R2 and CE (0.9999 and 0.9998) is the highest if it compared with the another BFGS Quasi-Newton model (6-3-1, 9-10-1 and 12-12-1). Based on the performance statistics value, MLRg, LM, SCG, BFGS and RBF suitable and accurately for prediction by modeling the non-linear complex behavior of suspended sediment responses to rainfall, water depth and discharge. The comparison between artificial neural network (ANN) and MLRg, the MLRg Model 2 accurately for to prediction suspended sediment discharge (kg
Xie, Jiaheng; Liu, Xiao; Dajun Zeng, Daniel
2018-01-01
Recent years have seen increased worldwide popularity of e-cigarette use. However, the risks of e-cigarettes are underexamined. Most e-cigarette adverse event studies have achieved low detection rates due to limited subject sample sizes in the experiments and surveys. Social media provides a large data repository of consumers' e-cigarette feedback and experiences, which are useful for e-cigarette safety surveillance. However, it is difficult to automatically interpret the informal and nontechnical consumer vocabulary about e-cigarettes in social media. This issue hinders the use of social media content for e-cigarette safety surveillance. Recent developments in deep neural network methods have shown promise for named entity extraction from noisy text. Motivated by these observations, we aimed to design a deep neural network approach to extract e-cigarette safety information in social media. Our deep neural language model utilizes word embedding as the representation of text input and recognizes named entity types with the state-of-the-art Bidirectional Long Short-Term Memory (Bi-LSTM) Recurrent Neural Network. Our Bi-LSTM model achieved the best performance compared to 3 baseline models, with a precision of 94.10%, a recall of 91.80%, and an F-measure of 92.94%. We identified 1591 unique adverse events and 9930 unique e-cigarette components (ie, chemicals, flavors, and devices) from our research testbed. Although the conditional random field baseline model had slightly better precision than our approach, our Bi-LSTM model achieved much higher recall, resulting in the best F-measure. Our method can be generalized to extract medical concepts from social media for other medical applications. © The Author 2017. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com
Non-linear auto-regressive models for cross-frequency coupling in neural time series
Tallot, Lucille; Grabot, Laetitia; Doyère, Valérie; Grenier, Yves; Gramfort, Alexandre
2017-01-01
We address the issue of reliably detecting and quantifying cross-frequency coupling (CFC) in neural time series. Based on non-linear auto-regressive models, the proposed method provides a generative and parametric model of the time-varying spectral content of the signals. As this method models the entire spectrum simultaneously, it avoids the pitfalls related to incorrect filtering or the use of the Hilbert transform on wide-band signals. As the model is probabilistic, it also provides a score of the model “goodness of fit” via the likelihood, enabling easy and legitimate model selection and parameter comparison; this data-driven feature is unique to our model-based approach. Using three datasets obtained with invasive neurophysiological recordings in humans and rodents, we demonstrate that these models are able to replicate previous results obtained with other metrics, but also reveal new insights such as the influence of the amplitude of the slow oscillation. Using simulations, we demonstrate that our parametric method can reveal neural couplings with shorter signals than non-parametric methods. We also show how the likelihood can be used to find optimal filtering parameters, suggesting new properties on the spectrum of the driving signal, but also to estimate the optimal delay between the coupled signals, enabling a directionality estimation in the coupling. PMID:29227989
Hariharan, M; Chee, Lim Sin; Yaacob, Sazali
2012-06-01
Acoustic analysis of infant cry signals has been proven to be an excellent tool in the area of automatic detection of pathological status of an infant. This paper investigates the application of parameter weighting for linear prediction cepstral coefficients (LPCCs) to provide the robust representation of infant cry signals. Three classes of infant cry signals were considered such as normal cry signals, cry signals from deaf babies and babies with asphyxia. A Probabilistic Neural Network (PNN) is suggested to classify the infant cry signals into normal and pathological cries. PNN is trained with different spread factor or smoothing parameter to obtain better classification accuracy. The experimental results demonstrate that the suggested features and classification algorithms give very promising classification accuracy of above 98% and it expounds that the suggested method can be used to help medical professionals for diagnosing pathological status of an infant from cry signals.
Tutubalina, Elena; Nikolenko, Sergey
2017-01-01
Adverse drug reactions (ADRs) are an essential part of the analysis of drug use, measuring drug use benefits, and making policy decisions. Traditional channels for identifying ADRs are reliable but very slow and only produce a small amount of data. Text reviews, either on specialized web sites or in general-purpose social networks, may lead to a data source of unprecedented size, but identifying ADRs in free-form text is a challenging natural language processing problem. In this work, we propose a novel model for this problem, uniting recurrent neural architectures and conditional random fields. We evaluate our model with a comprehensive experimental study, showing improvements over state-of-the-art methods of ADR extraction.
Directory of Open Access Journals (Sweden)
Elena Tutubalina
2017-01-01
Full Text Available Adverse drug reactions (ADRs are an essential part of the analysis of drug use, measuring drug use benefits, and making policy decisions. Traditional channels for identifying ADRs are reliable but very slow and only produce a small amount of data. Text reviews, either on specialized web sites or in general-purpose social networks, may lead to a data source of unprecedented size, but identifying ADRs in free-form text is a challenging natural language processing problem. In this work, we propose a novel model for this problem, uniting recurrent neural architectures and conditional random fields. We evaluate our model with a comprehensive experimental study, showing improvements over state-of-the-art methods of ADR extraction.
DEFF Research Database (Denmark)
Capaday, Charles; Ethier, C; Brizzi, L
2009-01-01
and functional significance of the intrinsic horizontal connections between neurons in the motor cortex (MCx) remain to be clarified. To further elucidate the nature of this intracortical connectivity pattern, experiments were done on the MCx of three cats. The anterograde tracer biocytin was ejected......Capaday C, Ethier C, Brizzi L, Sik A, van Vreeswijk C, Gingras D. On the nature of the intrinsic connectivity of the cat motor cortex: evidence for a recurrent neural network topology. J Neurophysiol 102: 2131-2141, 2009. First published July 22, 2009; doi: 10.1152/jn.91319.2008. The details...... iontophoretically in layers II, III, and V. Some 30-50 neurons within a radius of similar to 250 mu m were thus stained. The functional output of the motor cortical point at which biocytin was injected, and of the surrounding points, was identified by microstimulation and electromyographic recordings. The axonal...
Mioulet, L.; Bideault, G.; Chatelain, C.; Paquet, T.; Brunessaux, S.
2015-01-01
The BLSTM-CTC is a novel recurrent neural network architecture that has outperformed previous state of the art algorithms in tasks such as speech recognition or handwriting recognition. It has the ability to process long term dependencies in temporal signals in order to label unsegmented data. This paper describes different ways of combining features using a BLSTM-CTC architecture. Not only do we explore the low level combination (feature space combination) but we also explore high level combination (decoding combination) and mid-level (internal system representation combination). The results are compared on the RIMES word database. Our results show that the low level combination works best, thanks to the powerful data modeling of the LSTM neurons.
Rossi, A.; Montefoschi, F.; Rizzo, A.; Diligenti, M.; Festucci, C.
2017-10-01
Machine Learning applied to Automatic Audio Surveillance has been attracting increasing attention in recent years. In spite of several investigations based on a large number of different approaches, little attention had been paid to the environmental temporal evolution of the input signal. In this work, we propose an exploration in this direction comparing the temporal correlations extracted at the feature level with the one learned by a representational structure. To this aim we analysed the prediction performances of a Recurrent Neural Network architecture varying the length of the processed input sequence and the size of the time window used in the feature extraction. Results corroborated the hypothesis that sequential models work better when dealing with data characterized by temporal order. However, so far the optimization of the temporal dimension remains an open issue.
A Study of Recurrent and Convolutional Neural Networks in the Native Language Identification Task
Werfelmann, Robert
2018-05-24
Native Language Identification (NLI) is the task of predicting the native language of an author from their text written in a second language. The idea is to find writing habits that transfer from an author’s native language to their second language. Many approaches to this task have been studied, from simple word frequency analysis, to analyzing grammatical and spelling mistakes to find patterns and traits that are common between different authors of the same native language. This can be a very complex task, depending on the native language and the proficiency of the author’s second language. The most common approach that has seen very good results is based on the usage of n-gram features of words and characters. In this thesis, we attempt to extract lexical, grammatical, and semantic features from the sentences of non-native English essays using neural networks. The training and testing data was obtained from a large corpus of publicly available essays written by authors of several countries around the world. The neural network models consisted of Long Short-Term Memory and Convolutional networks using the sentences of each document as the input. Additional statistical features were generated from the text to complement the predictions of the neural networks, which were then used as feature inputs to a Support Vector Machine, making the final prediction. Results show that Long Short-Term Memory neural network can improve performance over a naive bag of words approach, but with a much smaller feature set. With more fine-tuning of neural network hyperparameters, these results will likely improve significantly.
Deep Recurrent Neural Networks for Product Attribute Extraction in eCommerce
Majumder, Bodhisattwa Prasad; Subramanian, Aditya; Krishnan, Abhinandan; Gandhi, Shreyansh; More, Ajinkya
2018-01-01
Extracting accurate attribute qualities from product titles is a vital component in delivering eCommerce customers with a rewarding online shopping experience via an enriched faceted search. We demonstrate the potential of Deep Recurrent Networks in this domain, primarily models such as Bidirectional LSTMs and Bidirectional LSTM-CRF with or without an attention mechanism. These have improved overall F1 scores, as compared to the previous benchmarks (More et al.) by at least 0.0391, showcasing...
Quang, Daniel; Xie, Xiaohui
2016-06-20
Modeling the properties and functions of DNA sequences is an important, but challenging task in the broad field of genomics. This task is particularly difficult for non-coding DNA, the vast majority of which is still poorly understood in terms of function. A powerful predictive model for the function of non-coding DNA can have enormous benefit for both basic science and translational research because over 98% of the human genome is non-coding and 93% of disease-associated variants lie in these regions. To address this need, we propose DanQ, a novel hybrid convolutional and bi-directional long short-term memory recurrent neural network framework for predicting non-coding function de novo from sequence. In the DanQ model, the convolution layer captures regulatory motifs, while the recurrent layer captures long-term dependencies between the motifs in order to learn a regulatory 'grammar' to improve predictions. DanQ improves considerably upon other models across several metrics. For some regulatory markers, DanQ can achieve over a 50% relative improvement in the area under the precision-recall curve metric compared to related models. We have made the source code available at the github repository http://github.com/uci-cbcl/DanQ. © The Author(s) 2016. Published by Oxford University Press on behalf of Nucleic Acids Research.
Khodabakhshi, Mohammad Bagher; Moradi, Mohammad Hassan
2017-05-01
The respiratory system dynamic is of high significance when it comes to the detection of lung abnormalities, which highlights the importance of presenting a reliable model for it. In this paper, we introduce a novel dynamic modelling method for the characterization of the lung sounds (LS), based on the attractor recurrent neural network (ARNN). The ARNN structure allows the development of an effective LS model. Additionally, it has the capability to reproduce the distinctive features of the lung sounds using its formed attractors. Furthermore, a novel ARNN topology based on fuzzy functions (FFs-ARNN) is developed. Given the utility of the recurrent quantification analysis (RQA) as a tool to assess the nature of complex systems, it was used to evaluate the performance of both the ARNN and the FFs-ARNN models. The experimental results demonstrate the effectiveness of the proposed approaches for multichannel LS analysis. In particular, a classification accuracy of 91% was achieved using FFs-ARNN with sequences of RQA features. Copyright © 2017 Elsevier Ltd. All rights reserved.
Hayashi, Hideaki; Shima, Keisuke; Shibanoki, Taro; Kurita, Yuichi; Tsuji, Toshio
2013-01-01
This paper outlines a probabilistic neural network developed on the basis of time-series discriminant component analysis (TSDCA) that can be used to classify high-dimensional time-series patterns. TSDCA involves the compression of high-dimensional time series into a lower-dimensional space using a set of orthogonal transformations and the calculation of posterior probabilities based on a continuous-density hidden Markov model that incorporates a Gaussian mixture model expressed in the reduced-dimensional space. The analysis can be incorporated into a neural network so that parameters can be obtained appropriately as network coefficients according to backpropagation-through-time-based training algorithm. The network is considered to enable high-accuracy classification of high-dimensional time-series patterns and to reduce the computation time taken for network training. In the experiments conducted during the study, the validity of the proposed network was demonstrated for EEG signals.
Song, Yongli; Makarov, Valeri A; Velarde, Manuel G
2009-08-01
A model of time-delay recurrently coupled spatially segregated neural assemblies is here proposed. We show that it operates like some of the hierarchical architectures of the brain. Each assembly is a neural network with no delay in the local couplings between the units. The delay appears in the long range feedforward and feedback inter-assemblies communications. Bifurcation analysis of a simple four-units system in the autonomous case shows the richness of the dynamical behaviors in a biophysically plausible parameter region. We find oscillatory multistability, hysteresis, and stability switches of the rest state provoked by the time delay. Then we investigate the spatio-temporal patterns of bifurcating periodic solutions by using the symmetric local Hopf bifurcation theory of delay differential equations and derive the equation describing the flow on the center manifold that enables us determining the direction of Hopf bifurcations and stability of the bifurcating periodic orbits. We also discuss computational properties of the system due to the delay when an external drive of the network mimicks external sensory input.
Directory of Open Access Journals (Sweden)
Run Min HOU
2014-05-01
Full Text Available To control the nonlinearity, widespread variations in loads and time varying characteristic of the high power ac servo system, the modeling and control techniques are studied here. A self-recurrent wavelet neural network (SRWNN modeling scheme is proposed, which successfully addresses the issue of the traditional wavelet neural network easily falling into local optimum, and significantly improves the network approximation capability and convergence rate. The control scheme of a SRWNN based on fuzzy compensation is expected. Gradient information is provided in real time for the controller by using a SRWNN identifier, so as to ensure that the learning and adjusting function of the controller of the SRWNN operate well, and fuzzy compensation control is applied to improve rapidity and accuracy of the entire system. Then the Lyapunov function is utilized to judge the stability of the system. The experimental analysis and comparisons with other modeling and control methods, it is clearly shown that the validities of the proposed modeling scheme and control scheme are effective.
Use of Recurrent Neural Networks for Strategic Data Mining of Sales
Vadhavkar, Sanjeev; Shanmugasundaram, Jayavel; Gupta, Amar; Prasad, M.V. Nagendra
2002-01-01
An increasing number of organizations are involved in the development of strategic information systems for effective linkages with their suppliers, customers, and other channel partners involved in transportation, distribution, warehousing and maintenance activities. An efficient inter-organizational inventory management system based on data mining techniques is a significant step in this direction. This paper discusses the use of neural network based data mining and knowledge discovery techn...
Cui, Zhiyong; Ke, Ruimin; Wang, Yinhai
2018-01-01
Short-term traffic forecasting based on deep learning methods, especially long short-term memory (LSTM) neural networks, has received much attention in recent years. However, the potential of deep learning methods in traffic forecasting has not yet fully been exploited in terms of the depth of the model architecture, the spatial scale of the prediction area, and the predictive power of spatial-temporal data. In this paper, a deep stacked bidirectional and unidirectional LSTM (SBU- LSTM) neura...
C-RNN-GAN: Continuous recurrent neural networks with adversarial training
Mogren, Olof
2016-01-01
Generative adversarial networks have been proposed as a way of efficiently training deep generative neural networks. We propose a generative adversarial model that works on continuous sequential data, and apply it by training it on a collection of classical music. We conclude that it generates music that sounds better and better as the model is trained, report statistics on generated music, and let the reader judge the quality by downloading the generated songs.
Learning to Recognize Actions From Limited Training Examples Using a Recurrent Spiking Neural Model
Panda, Priyadarshini; Srinivasa, Narayan
2018-01-01
A fundamental challenge in machine learning today is to build a model that can learn from few examples. Here, we describe a reservoir based spiking neural model for learning to recognize actions with a limited number of labeled videos. First, we propose a novel encoding, inspired by how microsaccades influence visual perception, to extract spike information from raw video data while preserving the temporal correlation across different frames. Using this encoding, we show that the reservoir generalizes its rich dynamical activity toward signature action/movements enabling it to learn from few training examples. We evaluate our approach on the UCF-101 dataset. Our experiments demonstrate that our proposed reservoir achieves 81.3/87% Top-1/Top-5 accuracy, respectively, on the 101-class data while requiring just 8 video examples per class for training. Our results establish a new benchmark for action recognition from limited video examples for spiking neural models while yielding competitive accuracy with respect to state-of-the-art non-spiking neural models. PMID:29551962
Buesing, Lars; Bill, Johannes; Nessler, Bernhard; Maass, Wolfgang
2011-11-01
The organization of computations in networks of spiking neurons in the brain is still largely unknown, in particular in view of the inherently stochastic features of their firing activity and the experimentally observed trial-to-trial variability of neural systems in the brain. In principle there exists a powerful computational framework for stochastic computations, probabilistic inference by sampling, which can explain a large number of macroscopic experimental data in neuroscience and cognitive science. But it has turned out to be surprisingly difficult to create a link between these abstract models for stochastic computations and more detailed models of the dynamics of networks of spiking neurons. Here we create such a link and show that under some conditions the stochastic firing activity of networks of spiking neurons can be interpreted as probabilistic inference via Markov chain Monte Carlo (MCMC) sampling. Since common methods for MCMC sampling in distributed systems, such as Gibbs sampling, are inconsistent with the dynamics of spiking neurons, we introduce a different approach based on non-reversible Markov chains that is able to reflect inherent temporal processes of spiking neuronal activity through a suitable choice of random variables. We propose a neural network model and show by a rigorous theoretical analysis that its neural activity implements MCMC sampling of a given distribution, both for the case of discrete and continuous time. This provides a step towards closing the gap between abstract functional models of cortical computation and more detailed models of networks of spiking neurons.
International Nuclear Information System (INIS)
Shao Hai-Jian; Cai Guo-Liang; Wang Hao-Xiang
2010-01-01
In this study, a successful linear matrix inequality approach is used to analyse a non-parameter perturbation of multi-delay Hopfield neural network by constructing an appropriate Lyapunov-Krasovskii functional. This paper presents the comprehensive discussion of the approach and also extensive applications
Recurrent-neural-network-based Boolean factor analysis and its application to word clustering.
Frolov, Alexander A; Husek, Dusan; Polyakov, Pavel Yu
2009-07-01
The objective of this paper is to introduce a neural-network-based algorithm for word clustering as an extension of the neural-network-based Boolean factor analysis algorithm (Frolov , 2007). It is shown that this extended algorithm supports even the more complex model of signals that are supposed to be related to textual documents. It is hypothesized that every topic in textual data is characterized by a set of words which coherently appear in documents dedicated to a given topic. The appearance of each word in a document is coded by the activity of a particular neuron. In accordance with the Hebbian learning rule implemented in the network, sets of coherently appearing words (treated as factors) create tightly connected groups of neurons, hence, revealing them as attractors of the network dynamics. The found factors are eliminated from the network memory by the Hebbian unlearning rule facilitating the search of other factors. Topics related to the found sets of words can be identified based on the words' semantics. To make the method complete, a special technique based on a Bayesian procedure has been developed for the following purposes: first, to provide a complete description of factors in terms of component probability, and second, to enhance the accuracy of classification of signals to determine whether it contains the factor. Since it is assumed that every word may possibly contribute to several topics, the proposed method might be related to the method of fuzzy clustering. In this paper, we show that the results of Boolean factor analysis and fuzzy clustering are not contradictory, but complementary. To demonstrate the capabilities of this attempt, the method is applied to two types of textual data on neural networks in two different languages. The obtained topics and corresponding words are at a good level of agreement despite the fact that identical topics in Russian and English conferences contain different sets of keywords.
International Nuclear Information System (INIS)
Fiorenza, Alberto; Vincenzi, Giovanni
2011-01-01
Research highlights: → We prove a result true for all linear homogeneous recurrences with constant coefficients. → As a corollary of our results we immediately get the celebrated Poincare' theorem. → The limit of the ratio of adjacent terms is characterized as the unique leading root of the characteristic polynomial. → The Golden Ratio, Kepler limit of the classical Fibonacci sequence, is the unique leading root. → The Kepler limit may differ from the unique root of maximum modulus and multiplicity. - Abstract: For complex linear homogeneous recursive sequences with constant coefficients we find a necessary and sufficient condition for the existence of the limit of the ratio of consecutive terms. The result can be applied even if the characteristic polynomial has not necessarily roots with modulus pairwise distinct, as in the celebrated Poincare's theorem. In case of existence, we characterize the limit as a particular root of the characteristic polynomial, which depends on the initial conditions and that is not necessarily the unique root with maximum modulus and multiplicity. The result extends to a quite general context the way used to find the Golden mean as limit of ratio of consecutive terms of the classical Fibonacci sequence.
Directory of Open Access Journals (Sweden)
Rachid Darnag
2017-02-01
Full Text Available Support vector machines (SVM represent one of the most promising Machine Learning (ML tools that can be applied to develop a predictive quantitative structure–activity relationship (QSAR models using molecular descriptors. Multiple linear regression (MLR and artificial neural networks (ANNs were also utilized to construct quantitative linear and non linear models to compare with the results obtained by SVM. The prediction results are in good agreement with the experimental value of HIV activity; also, the results reveal the superiority of the SVM over MLR and ANN model. The contribution of each descriptor to the structure–activity relationships was evaluated.
Learning in fully recurrent neural networks by approaching tangent planes to constraint surfaces.
May, P; Zhou, E; Lee, C W
2012-10-01
In this paper we present a new variant of the online real time recurrent learning algorithm proposed by Williams and Zipser (1989). Whilst the original algorithm utilises gradient information to guide the search towards the minimum training error, it is very slow in most applications and often gets stuck in local minima of the search space. It is also sensitive to the choice of learning rate and requires careful tuning. The new variant adjusts weights by moving to the tangent planes to constraint surfaces. It is simple to implement and requires no parameters to be set manually. Experimental results show that this new algorithm gives significantly faster convergence whilst avoiding problems like local minima. Copyright © 2012 Elsevier Ltd. All rights reserved.
Modeling the dynamics of human brain activity with recurrent neural networks
Güçlü, U.; Gerven, M.A.J. van
2017-01-01
Encoding models are used for predicting brain activity in response to sensory stimuli with the objective of elucidating how sensory information is represented in the brain. Encoding models typically comprise a nonlinear transformation of stimuli to features (feature model) and a linear convolution
DEFF Research Database (Denmark)
Kooths, Stefan; Mitze, Timo Friedel; Ringhut, Eric
2004-01-01
This paper compares the predictive power of linear econometric and non-linear computational models for forecasting the inflation rate in the European Monetary Union (EMU). Various models of both types are developed using different monetary and real activity indicators. They are compared according...
Data Normalization to Accelerate Training for Linear Neural Net to Predict Tropical Cyclone Tracks
Directory of Open Access Journals (Sweden)
Jian Jin
2015-01-01
Full Text Available When pure linear neural network (PLNN is used to predict tropical cyclone tracks (TCTs in South China Sea, whether the data is normalized or not greatly affects the training process. In this paper, min.-max. method and normal distribution method, instead of standard normal distribution, are applied to TCT data before modeling. We propose the experimental schemes in which, with min.-max. method, the min.-max. value pair of each variable is mapped to (−1, 1 and (0, 1; with normal distribution method, each variable’s mean and standard deviation pair is set to (0, 1 and (100, 1. We present the following results: (1 data scaled to the similar intervals have similar effects, no matter the use of min.-max. or normal distribution method; (2 mapping data to around 0 gains much faster training speed than mapping them to the intervals far away from 0 or using unnormalized raw data, although all of them can approach the same lower level after certain steps from their training error curves. This could be useful to decide data normalization method when PLNN is used individually.
An Ionospheric Index Model based on Linear Regression and Neural Network Approaches
Tshisaphungo, Mpho; McKinnell, Lee-Anne; Bosco Habarulema, John
2017-04-01
The ionosphere is well known to reflect radio wave signals in the high frequency (HF) band due to the present of electron and ions within the region. To optimise the use of long distance HF communications, it is important to understand the drivers of ionospheric storms and accurately predict the propagation conditions especially during disturbed days. This paper presents the development of an ionospheric storm-time index over the South African region for the application of HF communication users. The model will result into a valuable tool to measure the complex ionospheric behaviour in an operational space weather monitoring and forecasting environment. The development of an ionospheric storm-time index is based on a single ionosonde station data over Grahamstown (33.3°S,26.5°E), South Africa. Critical frequency of the F2 layer (foF2) measurements for a period 1996-2014 were considered for this study. The model was developed based on linear regression and neural network approaches. In this talk validation results for low, medium and high solar activity periods will be discussed to demonstrate model's performance.
Directory of Open Access Journals (Sweden)
Adi Syahputra
2014-03-01
Full Text Available Quantitative structure activity relationship (QSAR for 21 insecticides of phthalamides containing hydrazone (PCH was studied using multiple linear regression (MLR, principle component regression (PCR and artificial neural network (ANN. Five descriptors were included in the model for MLR and ANN analysis, and five latent variables obtained from principle component analysis (PCA were used in PCR analysis. Calculation of descriptors was performed using semi-empirical PM6 method. ANN analysis was found to be superior statistical technique compared to the other methods and gave a good correlation between descriptors and activity (r2 = 0.84. Based on the obtained model, we have successfully designed some new insecticides with higher predicted activity than those of previously synthesized compounds, e.g.2-(decalinecarbamoyl-5-chloro-N’-((5-methylthiophen-2-ylmethylene benzohydrazide, 2-(decalinecarbamoyl-5-chloro-N’-((thiophen-2-yl-methylene benzohydrazide and 2-(decaline carbamoyl-N’-(4-fluorobenzylidene-5-chlorobenzohydrazide with predicted log LC50 of 1.640, 1.672, and 1.769 respectively.
Directory of Open Access Journals (Sweden)
G. M. Behery
2009-01-01
Full Text Available This paper presents an automatic system of neural networks (NNs that has the ability to simulate and predict many of applied problems. The system architectures are automatically reorganized and the experimental process starts again, if the required performance is not reached. This processing is continued until the performance obtained. This system is first applied and tested on the two spiral problem; it shows that excellent generalization performance obtained by classifying all points of the two-spirals correctly. After that, it is applied and tested on the shear stress and the pressure drop problem across the short orifice die as a function of shear rate at different mean pressures for linear low-density polyethylene copolymer (LLDPE at 190∘C. The system shows a better agreement with an experimental data of the two cases: shear stress and pressure drop. The proposed system has been also designed to simulate other distributions not presented in the training set (predicted and matched them effectively.
Slowly evolving connectivity in recurrent neural networks: I. The extreme dilution regime
International Nuclear Information System (INIS)
Wemmenhove, B; Skantzos, N S; Coolen, A C C
2004-01-01
We study extremely diluted spin models of neural networks in which the connectivity evolves in time, although adiabatically slowly compared to the neurons, according to stochastic equations which on average aim to reduce frustration. The (fast) neurons and (slow) connectivity variables equilibrate separately, but at different temperatures. Our model is exactly solvable in equilibrium. We obtain phase diagrams upon making the condensed ansatz (i.e. recall of one pattern). These show that, as the connectivity temperature is lowered, the volume of the retrieval phase diverges and the fraction of mis-aligned spins is reduced. Still one always retains a region in the retrieval phase where recall states other than the one corresponding to the 'condensed' pattern are locally stable, so the associative memory character of our model is preserved
Schema generation in recurrent neural nets for intercepting a moving target.
Fleischer, Andreas G
2010-06-01
The grasping of a moving object requires the development of a motor strategy to anticipate the trajectory of the target and to compute an optimal course of interception. During the performance of perception-action cycles, a preprogrammed prototypical movement trajectory, a motor schema, may highly reduce the control load. Subjects were asked to hit a target that was moving along a circular path by means of a cursor. Randomized initial target positions and velocities were detected in the periphery of the eyes, resulting in a saccade toward the target. Even when the target disappeared, the eyes followed the target's anticipated course. The Gestalt of the trajectories was dependent on target velocity. The prediction capability of the motor schema was investigated by varying the visibility range of cursor and target. Motor schemata were determined to be of limited precision, and therefore visual feedback was continuously required to intercept the moving target. To intercept a target, the motor schema caused the hand to aim ahead and to adapt to the target trajectory. The control of cursor velocity determined the point of interception. From a modeling point of view, a neural network was developed that allowed the implementation of a motor schema interacting with feedback control in an iterative manner. The neural net of the Wilson type consists of an excitation-diffusion layer allowing the generation of a moving bubble. This activation bubble runs down an eye-centered motor schema and causes a planar arm model to move toward the target. A bubble provides local integration and straightening of the trajectory during repetitive moves. The schema adapts to task demands by learning and serves as forward controller. On the basis of these model considerations the principal problem of embedding motor schemata in generalized control strategies is discussed.
Krasilenko, Vladimir G.; Lazarev, Alexander A.; Nikitovich, Diana V.
2018-03-01
The biologically-motivated self-learning equivalence-convolutional recurrent-multilayer neural structures (BLM_SL_EC_RMNS) for fragments images clustering and recognition will be discussed. We shall consider these neural structures and their spatial-invariant equivalental models (SIEMs) based on proposed equivalent two-dimensional functions of image similarity and the corresponding matrix-matrix (or tensor) procedures using as basic operations of continuous logic and nonlinear processing. These SIEMs can simply describe the signals processing during the all training and recognition stages and they are suitable for unipolar-coding multilevel signals. The clustering efficiency in such models and their implementation depends on the discriminant properties of neural elements of hidden layers. Therefore, the main models and architecture parameters and characteristics depends on the applied types of non-linear processing and function used for image comparison or for adaptive-equivalent weighing of input patterns. We show that these SL_EC_RMNSs have several advantages, such as the self-study and self-identification of features and signs of the similarity of fragments, ability to clustering and recognize of image fragments with best efficiency and strong mutual correlation. The proposed combined with learning-recognition clustering method of fragments with regard to their structural features is suitable not only for binary, but also color images and combines self-learning and the formation of weight clustered matrix-patterns. Its model is constructed and designed on the basis of recursively continuous logic and nonlinear processing algorithms and to k-average method or method the winner takes all (WTA). The experimental results confirmed that fragments with a large numbers of elements may be clustered. For the first time the possibility of generalization of these models for space invariant case is shown. The experiment for an images of different dimensions (a reference
Hayashi, Hideaki; Shibanoki, Taro; Shima, Keisuke; Kurita, Yuichi; Tsuji, Toshio
2015-12-01
This paper proposes a probabilistic neural network (NN) developed on the basis of time-series discriminant component analysis (TSDCA) that can be used to classify high-dimensional time-series patterns. TSDCA involves the compression of high-dimensional time series into a lower dimensional space using a set of orthogonal transformations and the calculation of posterior probabilities based on a continuous-density hidden Markov model with a Gaussian mixture model expressed in the reduced-dimensional space. The analysis can be incorporated into an NN, which is named a time-series discriminant component network (TSDCN), so that parameters of dimensionality reduction and classification can be obtained simultaneously as network coefficients according to a backpropagation through time-based learning algorithm with the Lagrange multiplier method. The TSDCN is considered to enable high-accuracy classification of high-dimensional time-series patterns and to reduce the computation time taken for network training. The validity of the TSDCN is demonstrated for high-dimensional artificial data and electroencephalogram signals in the experiments conducted during the study.
Coding the presence of visual objects in a recurrent neural network of visual cortex.
Zwickel, Timm; Wachtler, Thomas; Eckhorn, Reinhard
2007-01-01
Before we can recognize a visual object, our visual system has to segregate it from its background. This requires a fast mechanism for establishing the presence and location of objects independently of their identity. Recently, border-ownership neurons were recorded in monkey visual cortex which might be involved in this task [Zhou, H., Friedmann, H., von der Heydt, R., 2000. Coding of border ownership in monkey visual cortex. J. Neurosci. 20 (17), 6594-6611]. In order to explain the basic mechanisms required for fast coding of object presence, we have developed a neural network model of visual cortex consisting of three stages. Feed-forward and lateral connections support coding of Gestalt properties, including similarity, good continuation, and convexity. Neurons of the highest area respond to the presence of an object and encode its position, invariant of its form. Feedback connections to the lowest area facilitate orientation detectors activated by contours belonging to potential objects, and thus generate the experimentally observed border-ownership property. This feedback control acts fast and significantly improves the figure-ground segregation required for the consecutive task of object recognition.
International Nuclear Information System (INIS)
Boroushaki, M.; Ghofrani, M.B.; Lucas, C.; Yazdanpanah, M.J.
2003-01-01
In the last decade, the intelligent control community has paid great attention to the topic of intelligent control systems for nuclear plants (core, steam generator...). Papers mostly used approximate and simple mathematical SISO (single-input-single-output) model of nuclear plants for testing and/or tuning of the control systems. They also tried to generalize theses models to a real MIMO (multi-input-multi-output) plant, while nuclear plants are typically of complex nonlinear and multivariable nature with high interactions between their state variables and therefore, many of these proposed intelligent control systems are not appropriate for real cases. In this paper, we designed an on-line intelligent core controller for load following operations, based on a heuristic control algorithm, using a valid and updatable recurrent neural network (RNN). We have used an accurate 3-dimensional core calculation code to represent the real plant and to train the RNN. The results of simulation show that this intelligent controller can control the reactor core during load following operations, using optimum control rod groups manoeuvre and variable overlapping strategy. This methodology represents a simple and reliable procedure for controlling other complex nonlinear MIMO plants, and may improve the responses, comparing to other control systems
Directory of Open Access Journals (Sweden)
YuKang Jia
2017-01-01
Full Text Available Long Short-Term Memory (LSTM is a kind of Recurrent Neural Networks (RNN relating to time series, which has achieved good performance in speech recogniton and image recognition. Long Short-Term Memory Projection (LSTMP is a variant of LSTM to further optimize speed and performance of LSTM by adding a projection layer. As LSTM and LSTMP have performed well in pattern recognition, in this paper, we combine them with Connectionist Temporal Classification (CTC to study piano’s continuous note recognition for robotics. Based on the Beijing Forestry University music library, we conduct experiments to show recognition rates and numbers of iterations of LSTM with a single layer, LSTMP with a single layer, and Deep LSTM (DLSTM, LSTM with multilayers. As a result, the single layer LSTMP proves performing much better than the single layer LSTM in both time and the recognition rate; that is, LSTMP has fewer parameters and therefore reduces the training time, and, moreover, benefiting from the projection layer, LSTMP has better performance, too. The best recognition rate of LSTMP is 99.8%. As for DLSTM, the recognition rate can reach 100% because of the effectiveness of the deep structure, but compared with the single layer LSTMP, DLSTM needs more training time.
Jimeno Yepes, Antonio
2017-09-01
Word sense disambiguation helps identifying the proper sense of ambiguous words in text. With large terminologies such as the UMLS Metathesaurus ambiguities appear and highly effective disambiguation methods are required. Supervised learning algorithm methods are used as one of the approaches to perform disambiguation. Features extracted from the context of an ambiguous word are used to identify the proper sense of such a word. The type of features have an impact on machine learning methods, thus affect disambiguation performance. In this work, we have evaluated several types of features derived from the context of the ambiguous word and we have explored as well more global features derived from MEDLINE using word embeddings. Results show that word embeddings improve the performance of more traditional features and allow as well using recurrent neural network classifiers based on Long-Short Term Memory (LSTM) nodes. The combination of unigrams and word embeddings with an SVM sets a new state of the art performance with a macro accuracy of 95.97 in the MSH WSD data set. Copyright © 2017 Elsevier Inc. All rights reserved.
International Nuclear Information System (INIS)
You, Seung Han; Hahn, Jin Oh
2012-01-01
By virtue of its ease of operation compared with its conventional manual counterpart, automatic transmissions are commonly used as automotive power transmission control system in today's passenger cars. In accordance with this trend, research efforts on closed-loop automatic transmission controls have been extensively carried out to improve ride quality and fuel economy. State-of-the-art power transmission control algorithms may have limitations in performance because they rely on the steady-state characteristics of the hydraulic actuator rather than fully exploit its dynamic characteristics. Since the ultimate viability of closed-loop power transmission control is dominated by precise pressure control at the level of hydraulic actuator, closed-loop control can potentially attain superior efficacy in case the hydraulic actuator can be easily incorporated into model-based observer/controller design. In this paper, we propose to use a recurrent neural network (RNN) to establish a nonlinear empirical model of a cascade hydraulic actuator in a passenger car automatic transmission, which has potential to be easily incorporated in designing observers and controllers. Experimental analysis is performed to grasp key system characteristics, based on which a nonlinear system identification procedure is carried out. Extensive experimental validation of the established model suggests that it has superb one-step-ahead prediction capability over appropriate frequency range, making it an attractive approach for model-based observer/controller design applications in automotive systems
Xiao, Lin; Zhang, Yongsheng; Liao, Bolin; Zhang, Zhijun; Ding, Lei; Jin, Long
2017-01-01
A dual-robot system is a robotic device composed of two robot arms. To eliminate the joint-angle drift and prevent the occurrence of high joint velocity, a velocity-level bi-criteria optimization scheme, which includes two criteria (i.e., the minimum velocity norm and the repetitive motion), is proposed and investigated for coordinated path tracking of dual robot manipulators. Specifically, to realize the coordinated path tracking of dual robot manipulators, two subschemes are first presented for the left and right robot manipulators. After that, such two subschemes are reformulated as two general quadratic programs (QPs), which can be formulated as one unified QP. A recurrent neural network (RNN) is thus presented to solve effectively the unified QP problem. At last, computer simulation results based on a dual three-link planar manipulator further validate the feasibility and the efficacy of the velocity-level optimization scheme for coordinated path tracking using the recurrent neural network.
Wang, J; Wang, F; Liu, Y; Xu, J; Lin, H; Jia, B; Zuo, W; Jiang, Y; Hu, L; Lin, F
2016-01-01
Overweight individuals are at higher risk for developing type II diabetes than the general population. We conducted this study to analyze the correlation between blood glucose and biochemical parameters, and developed a blood glucose prediction model tailored to overweight patients. A total of 346 overweight Chinese people patients ages 18-81 years were involved in this study. Their levels of fasting glucose (fs-GLU), blood lipids, and hepatic and renal functions were measured and analyzed by multiple linear regression (MLR). Based the MLR results, we developed a back propagation artificial neural network (BP-ANN) model by selecting tansig as the transfer function of the hidden layers nodes, and purelin for the output layer nodes, with training goal of 0.5×10(-5). There was significant correlation between fs-GLU with age, BMI, and blood biochemical indexes (P<0.05). The results of MLR analysis indicated that age, fasting alanine transaminase (fs-ALT), blood urea nitrogen (fs-BUN), total protein (fs-TP), uric acid (fs-BUN), and BMI are 6 independent variables related to fs-GLU. Based on these parameters, the BP-ANN model was performed well and reached high prediction accuracy when training 1 000 epoch (R=0.9987). The level of fs-GLU was predictable using the proposed BP-ANN model based on 6 related parameters (age, fs-ALT, fs-BUN, fs-TP, fs-UA and BMI) in overweight patients. © Georg Thieme Verlag KG Stuttgart · New York.
Yakubu, A.; Oluremi, O. I. A.; Ekpo, E. I.
2018-03-01
There is an increasing use of robust analytical algorithms in the prediction of heat stress. The present investigation therefore, was carried out to forecast heat stress index (HSI) in Sasso laying hens. One hundred and sixty seven records on the thermo-physiological parameters of the birds were utilized. They were reared on deep litter and battery cage systems. Data were collected when the birds were 42- and 52-week of age. The independent variables fitted were housing system, age of birds, rectal temperature (RT), pulse rate (PR), and respiratory rate (RR). The response variable was HSI. Data were analyzed using automatic linear modeling (ALM) and artificial neural network (ANN) procedures. The ALM model building method involved Forward Stepwise using the F Statistic criterion. As regards ANN, multilayer perceptron (MLP) with back-propagation network was used. The ANN network was trained with 90% of the data set while 10% were dedicated to testing for model validation. RR and PR were the two parameters of utmost importance in the prediction of HSI. However, the fractional importance of RR was higher than that of PR in both ALM (0.947 versus 0.053) and ANN (0.677 versus 0.274) models. The two models also predicted HSI effectively with high degree of accuracy [r = 0.980, R 2 = 0.961, adjusted R 2 = 0.961, and RMSE = 0.05168 (ALM); r = 0.983, R 2 = 0.966; adjusted R 2 = 0.966, and RMSE = 0.04806 (ANN)]. The present information may be exploited in the development of a heat stress chart based largely on RR. This may aid detection of thermal discomfort in a poultry house under tropical and subtropical conditions.
Everson, Howard T.; And Others
This paper explores the feasibility of neural computing methods such as artificial neural networks (ANNs) and abductory induction mechanisms (AIM) for use in educational measurement. ANNs and AIMS methods are contrasted with more traditional statistical techniques, such as multiple regression and discriminant function analyses, for making…
Siri, Benoît; Berry, Hugues; Cessac, Bruno; Delord, Bruno; Quoy, Mathias
2008-12-01
We present a mathematical analysis of the effects of Hebbian learning in random recurrent neural networks, with a generic Hebbian learning rule, including passive forgetting and different timescales, for neuronal activity and learning dynamics. Previous numerical work has reported that Hebbian learning drives the system from chaos to a steady state through a sequence of bifurcations. Here, we interpret these results mathematically and show that these effects, involving a complex coupling between neuronal dynamics and synaptic graph structure, can be analyzed using Jacobian matrices, which introduce both a structural and a dynamical point of view on neural network evolution. Furthermore, we show that sensitivity to a learned pattern is maximal when the largest Lyapunov exponent is close to 0. We discuss how neural networks may take advantage of this regime of high functional interest.
Nie, Xiaobing; Zheng, Wei Xing
2015-05-01
This paper is concerned with the problem of coexistence and dynamical behaviors of multiple equilibrium points for neural networks with discontinuous non-monotonic piecewise linear activation functions and time-varying delays. The fixed point theorem and other analytical tools are used to develop certain sufficient conditions that ensure that the n-dimensional discontinuous neural networks with time-varying delays can have at least 5(n) equilibrium points, 3(n) of which are locally stable and the others are unstable. The importance of the derived results is that it reveals that the discontinuous neural networks can have greater storage capacity than the continuous ones. Moreover, different from the existing results on multistability of neural networks with discontinuous activation functions, the 3(n) locally stable equilibrium points obtained in this paper are located in not only saturated regions, but also unsaturated regions, due to the non-monotonic structure of discontinuous activation functions. A numerical simulation study is conducted to illustrate and support the derived theoretical results. Copyright © 2015 Elsevier Ltd. All rights reserved.
A new method to estimate parameters of linear compartmental models using artificial neural networks
International Nuclear Information System (INIS)
Gambhir, Sanjiv S.; Keppenne, Christian L.; Phelps, Michael E.; Banerjee, Pranab K.
1998-01-01
At present, the preferred tool for parameter estimation in compartmental analysis is an iterative procedure; weighted nonlinear regression. For a large number of applications, observed data can be fitted to sums of exponentials whose parameters are directly related to the rate constants/coefficients of the compartmental models. Since weighted nonlinear regression often has to be repeated for many different data sets, the process of fitting data from compartmental systems can be very time consuming. Furthermore the minimization routine often converges to a local (as opposed to global) minimum. In this paper, we examine the possibility of using artificial neural networks instead of weighted nonlinear regression in order to estimate model parameters. We train simple feed-forward neural networks to produce as outputs the parameter values of a given model when kinetic data are fed to the networks' input layer. The artificial neural networks produce unbiased estimates and are orders of magnitude faster than regression algorithms. At noise levels typical of many real applications, the neural networks are found to produce lower variance estimates than weighted nonlinear regression in the estimation of parameters from mono- and biexponential models. These results are primarily due to the inability of weighted nonlinear regression to converge. These results establish that artificial neural networks are powerful tools for estimating parameters for simple compartmental models. (author)
Güntürkün, Rüştü
2010-08-01
In this study, Elman recurrent neural networks have been defined by using conjugate gradient algorithm in order to determine the depth of anesthesia in the continuation stage of the anesthesia and to estimate the amount of medicine to be applied at that moment. The feed forward neural networks are also used for comparison. The conjugate gradient algorithm is compared with back propagation (BP) for training of the neural Networks. The applied artificial neural network is composed of three layers, namely the input layer, the hidden layer and the output layer. The nonlinear activation function sigmoid (sigmoid function) has been used in the hidden layer and the output layer. EEG data has been recorded with Nihon Kohden 9200 brand 22-channel EEG device. The international 8-channel bipolar 10-20 montage system (8 TB-b system) has been used in assembling the recording electrodes. EEG data have been recorded by being sampled once in every 2 milliseconds. The artificial neural network has been designed so as to have 60 neurons in the input layer, 30 neurons in the hidden layer and 1 neuron in the output layer. The values of the power spectral density (PSD) of 10-second EEG segments which correspond to the 1-50 Hz frequency range; the ratio of the total power of PSD values of the EEG segment at that moment in the same range to the total of PSD values of EEG segment taken prior to the anesthesia.
浅野, 美代子; マーコ, ユー K.W.
2007-01-01
This paper introduces the hybrid approach of neural networks and linear regression model proposed by Asano and Tsubaki (2003). Neural networks are often credited with its superiority in data consistency whereas the linear regression model provides simple interpretation of the data enabling researchers to verify their hypotheses. The hybrid approach aims at combing the strengths of these two well-established statistical methods. A step-by-step procedure for performing the hybrid approach is pr...
Directory of Open Access Journals (Sweden)
Luis Gonzaga Baca Ruiz
2016-08-01
Full Text Available This paper addresses the problem of energy consumption prediction using neural networks over a set of public buildings. Since energy consumption in the public sector comprises a substantial share of overall consumption, the prediction of such consumption represents a decisive issue in the achievement of energy savings. In our experiments, we use the data provided by an energy consumption monitoring system in a compound of faculties and research centers at the University of Granada, and provide a methodology to predict future energy consumption using nonlinear autoregressive (NAR and the nonlinear autoregressive neural network with exogenous inputs (NARX, respectively. Results reveal that NAR and NARX neural networks are both suitable for performing energy consumption prediction, but also that exogenous data may help to improve the accuracy of predictions.
Ventura, Cristina; Latino, Diogo A R S; Martins, Filomena
2013-01-01
The performance of two QSAR methodologies, namely Multiple Linear Regressions (MLR) and Neural Networks (NN), towards the modeling and prediction of antitubercular activity was evaluated and compared. A data set of 173 potentially active compounds belonging to the hydrazide family and represented by 96 descriptors was analyzed. Models were built with Multiple Linear Regressions (MLR), single Feed-Forward Neural Networks (FFNNs), ensembles of FFNNs and Associative Neural Networks (AsNNs) using four different data sets and different types of descriptors. The predictive ability of the different techniques used were assessed and discussed on the basis of different validation criteria and results show in general a better performance of AsNNs in terms of learning ability and prediction of antitubercular behaviors when compared with all other methods. MLR have, however, the advantage of pinpointing the most relevant molecular characteristics responsible for the behavior of these compounds against Mycobacterium tuberculosis. The best results for the larger data set (94 compounds in training set and 18 in test set) were obtained with AsNNs using seven descriptors (R(2) of 0.874 and RMSE of 0.437 against R(2) of 0.845 and RMSE of 0.472 in MLRs, for test set). Counter-Propagation Neural Networks (CPNNs) were trained with the same data sets and descriptors. From the scrutiny of the weight levels in each CPNN and the information retrieved from MLRs, a rational design of potentially active compounds was attempted. Two new compounds were synthesized and tested against M. tuberculosis showing an activity close to that predicted by the majority of the models. Copyright © 2013 Elsevier Masson SAS. All rights reserved.
International Nuclear Information System (INIS)
Kim, Yong-il; Kim, Yong Joong; Paeng, Jin Chul; Cheon, Gi Jeong; Lee, Dong Soo; Chung, June-Key; Kang, Keon Wook
2017-01-01
18 F-Fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) has been investigated as a method to predict pancreatic cancer recurrence after pancreatic surgery. We evaluated the recently introduced heterogeneity indices of 18 F-FDG PET/CT used for predicting pancreatic cancer recurrence after surgery and compared them with current clinicopathologic and 18 F-FDG PET/CT parameters. A total of 93 pancreatic ductal adenocarcinoma patients (M:F = 60:33, mean age = 64.2 ± 9.1 years) who underwent preoperative 18 F-FDG PET/CT following pancreatic surgery were retrospectively enrolled. The standardized uptake values (SUVs) and tumor-to-background ratios (TBR) were measured on each 18 F-FDG PET/CT, as metabolic parameters. Metabolic tumor volume (MTV) and total lesion glycolysis (TLG) were examined as volumetric parameters. The coefficient of variance (heterogeneity index-1; SUVmean divided by the standard deviation) and linear regression slopes (heterogeneity index-2) of the MTV, according to SUV thresholds of 2.0, 2.5 and 3.0, were evaluated as heterogeneity indices. Predictive values of clinicopathologic and 18 F-FDG PET/CT parameters and heterogeneity indices were compared in terms of pancreatic cancer recurrence. Seventy patients (75.3%) showed recurrence after pancreatic cancer surgery (mean recurrence = 9.4 ± 8.4 months). Comparing the recurrence and no recurrence patients, all of the 18 F-FDG PET/CT parameters and heterogeneity indices demonstrated significant differences. In univariate Cox-regression analyses, MTV (P = 0.013), TLG (P = 0.007), and heterogeneity index-2 (P = 0.027) were significant. Among the clinicopathologic parameters, CA19-9 (P = 0.025) and venous invasion (P = 0.002) were selected as significant parameters. In multivariate Cox-regression analyses, MTV (P = 0.005), TLG (P = 0.004), and heterogeneity index-2 (P = 0.016) with venous invasion (P < 0.001, 0.001, and 0.001, respectively) demonstrated significant results
Directory of Open Access Journals (Sweden)
V. Rezan USLU
2010-01-01
Full Text Available Obtaining the inflation prediction is an important problem. Having this prediction accurately will lead to more accurate decisions. Various time series techniques have been used in the literature for inflation prediction. Recently, Artificial Neural Network (ANN is being preferred in the time series prediction problem due to its flexible modeling capacity. Artificial neural network can be applied easily to any time series since it does not require prior conditions such as a linear or curved specific model pattern, stationary and normal distribution. In this study, the predictions have been obtained using the feed forward and recurrent artificial neural network for the Consumer Price Index (CPI. A new combined forecast has been proposed based on ANN in which the ANN model predictions employed in analysis were used as data.
Hsu, Ching-Chi; Lin, Jinn; Chao, Ching-Kong
2011-12-01
Optimizing the orthopaedic screws can greatly improve their biomechanical performances. However, a methodical design optimization approach requires a long time to search the best design. Thus, the surrogate objective functions of the orthopaedic screws should be accurately developed. To our knowledge, there is no study to evaluate the strengths and limitations of the surrogate methods in developing the objective functions of the orthopaedic screws. Three-dimensional finite element models for both the tibial locking screws and the spinal pedicle screws were constructed and analyzed. Then, the learning data were prepared according to the arrangement of the Taguchi orthogonal array, and the verification data were selected with use of a randomized selection. Finally, the surrogate objective functions were developed by using either the multiple linear regression or the artificial neural network. The applicability and accuracy of those surrogate methods were evaluated and discussed. The multiple linear regression method could successfully construct the objective function of the tibial locking screws, but it failed to develop the objective function of the spinal pedicle screws. The artificial neural network method showed a greater capacity of prediction in developing the objective functions for the tibial locking screws and the spinal pedicle screws than the multiple linear regression method. The artificial neural network method may be a useful option for developing the objective functions of the orthopaedic screws with a greater structural complexity. The surrogate objective functions of the orthopaedic screws could effectively decrease the time and effort required for the design optimization process. Copyright Â© 2010 Elsevier Ireland Ltd. All rights reserved.
Neural network modeling of nonlinear systems based on Volterra series extension of a linear model
Soloway, Donald I.; Bialasiewicz, Jan T.
1992-01-01
A Volterra series approach was applied to the identification of nonlinear systems which are described by a neural network model. A procedure is outlined by which a mathematical model can be developed from experimental data obtained from the network structure. Applications of the results to the control of robotic systems are discussed.
Lo, Ching F.
1999-01-01
The integration of Radial Basis Function Networks and Back Propagation Neural Networks with the Multiple Linear Regression has been accomplished to map nonlinear response surfaces over a wide range of independent variables in the process of the Modem Design of Experiments. The integrated method is capable to estimate the precision intervals including confidence and predicted intervals. The power of the innovative method has been demonstrated by applying to a set of wind tunnel test data in construction of response surface and estimation of precision interval.
Directory of Open Access Journals (Sweden)
Adrian TOADER
2010-09-01
Full Text Available The paper was conceived in two parts. Part I, previously published in this journal, highlighted the main steps of adaptive output feedback control for non-affine uncertain systems, having a known relative degree. The main paradigm of this approach was the feedback linearization (dynamic inversion with neural network augmentation. Meanwhile, based on new contributions of the authors, a new paradigm, that of robust servomechanism problem solution, has been added to the controller architecture. The current Part II of the paper presents the validation of the controller hereby obtained by using the longitudinal channel of a hovering VTOL-type aircraft as mathematical model.
International Nuclear Information System (INIS)
Karthik Raja, U; Leelamani, A; Raja, R; Samidurai, R
2013-01-01
In this paper, the exponential stability for a class of stochastic neural networks with time-varying delays and impulsive effects is considered. By constructing suitable Lyapunov functionals and by using the linear matrix inequality optimization approach, we obtain sufficient delay-dependent criteria to ensure the exponential stability of stochastic neural networks with time-varying delays and impulses. Two numerical examples with simulation results are provided to illustrate the effectiveness of the obtained results over those already existing in the literature. (paper)
International Nuclear Information System (INIS)
Ghaffari, A.; Nikkhah Bahrami, M.; Mohammadzaheri, M.
2005-01-01
In this paper a new method for linear modeling of nonlinear systems is presented. The method is based on the design of an artificial neural network with two layers. The network is trained only according to the input-output data of the system. The weights of connections in this network, represents the coefficients of the transfer function. For systems with linear behavior the method of least square error represents the best linear model of the system. However, for nonlinear systems, such as some subsystems in power plants boilers LSE does not represent the best linear approximation of the system, necessarily. In this paper a new linear modeling method is presented and applied to some subsystems in a power plant boiler. Comparison between the transfer function obtained in this way and by least square error method,shows that the neural network method gives better linear models for these nonlinear systems
BRITS: Bidirectional Recurrent Imputation for Time Series
Cao, Wei; Wang, Dong; Li, Jian; Zhou, Hao; Li, Lei; Li, Yitan
2018-01-01
Time series are widely used as signals in many classification/regression tasks. It is ubiquitous that time series contains many missing values. Given multiple correlated time series data, how to fill in missing values and to predict their class labels? Existing imputation methods often impose strong assumptions of the underlying data generating process, such as linear dynamics in the state space. In this paper, we propose BRITS, a novel method based on recurrent neural networks for missing va...
Recurrent networks for wave forecasting
Digital Repository Service at National Institute of Oceanography (India)
Mandal, S.; Prabaharan, N.
, merchant vessel routing, nearshore construction, etc. more efficiently and safely. This paper presents an application of the Artificial Neural Network, namely Backpropagation Recurrent Neural Network (BRNN) with rprop update algorithm for wave forecasting...
International Nuclear Information System (INIS)
Jahandideh, Sepideh; Jahandideh, Samad; Asadabadi, Ebrahim Barzegari; Askarian, Mehrdad; Movahedi, Mohammad Mehdi; Hosseini, Somayyeh; Jahandideh, Mina
2009-01-01
Prediction of the amount of hospital waste production will be helpful in the storage, transportation and disposal of hospital waste management. Based on this fact, two predictor models including artificial neural networks (ANNs) and multiple linear regression (MLR) were applied to predict the rate of medical waste generation totally and in different types of sharp, infectious and general. In this study, a 5-fold cross-validation procedure on a database containing total of 50 hospitals of Fars province (Iran) were used to verify the performance of the models. Three performance measures including MAR, RMSE and R 2 were used to evaluate performance of models. The MLR as a conventional model obtained poor prediction performance measure values. However, MLR distinguished hospital capacity and bed occupancy as more significant parameters. On the other hand, ANNs as a more powerful model, which has not been introduced in predicting rate of medical waste generation, showed high performance measure values, especially 0.99 value of R 2 confirming the good fit of the data. Such satisfactory results could be attributed to the non-linear nature of ANNs in problem solving which provides the opportunity for relating independent variables to dependent ones non-linearly. In conclusion, the obtained results showed that our ANN-based model approach is very promising and may play a useful role in developing a better cost-effective strategy for waste management in future.
A High-Linearity Low-Noise Amplifier with Variable Bandwidth for Neural Recoding Systems
Yoshida, Takeshi; Sueishi, Katsuya; Iwata, Atsushi; Matsushita, Kojiro; Hirata, Masayuki; Suzuki, Takafumi
2011-04-01
This paper describes a low-noise amplifier with multiple adjustable parameters for neural recording applications. An adjustable pseudo-resistor implemented by cascade metal-oxide-silicon field-effect transistors (MOSFETs) is proposed to achieve low-signal distortion and wide variable bandwidth range. The amplifier has been implemented in 0.18 µm standard complementary metal-oxide-semiconductor (CMOS) process and occupies 0.09 mm2 on chip. The amplifier achieved a selectable voltage gain of 28 and 40 dB, variable bandwidth from 0.04 to 2.6 Hz, total harmonic distortion (THD) of 0.2% with 200 mV output swing, input referred noise of 2.5 µVrms over 0.1-100 Hz and 18.7 µW power consumption at a supply voltage of 1.8 V.
Chen, Xiaofeng; Song, Qiankun; Li, Zhongshan; Zhao, Zhenjiang; Liu, Yurong
2018-07-01
This paper addresses the problem of stability for continuous-time and discrete-time quaternion-valued neural networks (QVNNs) with linear threshold neurons. Applying the semidiscretization technique to the continuous-time QVNNs, the discrete-time analogs are obtained, which preserve the dynamical characteristics of their continuous-time counterparts. Via the plural decomposition method of quaternion, homeomorphic mapping theorem, as well as Lyapunov theorem, some sufficient conditions on the existence, uniqueness, and global asymptotical stability of the equilibrium point are derived for the continuous-time QVNNs and their discrete-time analogs, respectively. Furthermore, a uniform sufficient condition on the existence, uniqueness, and global asymptotical stability of the equilibrium point is obtained for both continuous-time QVNNs and their discrete-time version. Finally, two numerical examples are provided to substantiate the effectiveness of the proposed results.
Singh, S.; Jaishi, H. P.; Tiwari, R. P.; Tiwari, R. C.
2017-07-01
This paper reports the analysis of soil radon data recorded in the seismic zone-V, located in the northeastern part of India (latitude 23.73N, longitude 92.73E). Continuous measurements of soil-gas emission along Chite fault in Mizoram (India) were carried out with the replacement of solid-state nuclear track detectors at weekly interval. The present study was done for the period from March 2013 to May 2015 using LR-115 Type II detectors, manufactured by Kodak Pathe, France. In order to reduce the influence of meteorological parameters, statistical analysis tools such as multiple linear regression and artificial neural network have been used. Decrease in radon concentration was recorded prior to some earthquakes that occurred during the observation period. Some false anomalies were also recorded which may be attributed to the ongoing crustal deformation which was not major enough to produce an earthquake.
Yong, Yan Ling; Tan, Li Kuo; McLaughlin, Robert A.; Chee, Kok Han; Liew, Yih Miin
2017-12-01
Intravascular optical coherence tomography (OCT) is an optical imaging modality commonly used in the assessment of coronary artery diseases during percutaneous coronary intervention. Manual segmentation to assess luminal stenosis from OCT pullback scans is challenging and time consuming. We propose a linear-regression convolutional neural network to automatically perform vessel lumen segmentation, parameterized in terms of radial distances from the catheter centroid in polar space. Benchmarked against gold-standard manual segmentation, our proposed algorithm achieves average locational accuracy of the vessel wall of 22 microns, and 0.985 and 0.970 in Dice coefficient and Jaccard similarity index, respectively. The average absolute error of luminal area estimation is 1.38%. The processing rate is 40.6 ms per image, suggesting the potential to be incorporated into a clinical workflow and to provide quantitative assessment of vessel lumen in an intraoperative time frame.
Yong, Yan Ling; Tan, Li Kuo; McLaughlin, Robert A; Chee, Kok Han; Liew, Yih Miin
2017-12-01
Intravascular optical coherence tomography (OCT) is an optical imaging modality commonly used in the assessment of coronary artery diseases during percutaneous coronary intervention. Manual segmentation to assess luminal stenosis from OCT pullback scans is challenging and time consuming. We propose a linear-regression convolutional neural network to automatically perform vessel lumen segmentation, parameterized in terms of radial distances from the catheter centroid in polar space. Benchmarked against gold-standard manual segmentation, our proposed algorithm achieves average locational accuracy of the vessel wall of 22 microns, and 0.985 and 0.970 in Dice coefficient and Jaccard similarity index, respectively. The average absolute error of luminal area estimation is 1.38%. The processing rate is 40.6 ms per image, suggesting the potential to be incorporated into a clinical workflow and to provide quantitative assessment of vessel lumen in an intraoperative time frame. (2017) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE).
Integrating piecewise linear representation and ensemble neural network for stock price prediction
Asaduzzaman, Md.; Shahjahan, Md.; Ahmed, Fatema Johera; Islam, Md. Monirul; Murase, Kazuyuki
2014-01-01
Stock Prices are considered to be very dynamic and susceptible to quick changes because of the underlying nature of the financial domain, and in part because of the interchange between known parameters and unknown factors. Of late, several researchers have used Piecewise Linear Representation (PLR) to predict the stock market pricing. However, some improvements are needed to avoid the appropriate threshold of the trading decision, choosing the input index as well as improving the overall perf...
International Nuclear Information System (INIS)
Yoshioka, Masahiko
2002-01-01
We study associative memory neural networks of the Hodgkin-Huxley type of spiking neurons in which multiple periodic spatiotemporal patterns of spike timing are memorized as limit-cycle-type attractors. In encoding the spatiotemporal patterns, we assume the spike-timing-dependent synaptic plasticity with the asymmetric time window. Analysis for periodic solution of retrieval state reveals that if the area of the negative part of the time window is equivalent to the positive part, then crosstalk among encoded patterns vanishes. Phase transition due to the loss of the stability of periodic solution is observed when we assume fast α function for direct interaction among neurons. In order to evaluate the critical point of this phase transition, we employ Floquet theory in which the stability problem of the infinite number of spiking neurons interacting with α function is reduced to the eigenvalue problem with the finite size of matrix. Numerical integration of the single-body dynamics yields the explicit value of the matrix, which enables us to determine the critical point of the phase transition with a high degree of precision
2017-11-07
Adult Anaplastic Astrocytoma; Recurrent Grade III Glioma; Recurrent Grade IV Glioma; Adult Anaplastic Oligodendroglioma; Adult Brain Tumor; Adult Giant Cell Glioblastoma; Adult Glioblastoma; Adult Gliosarcoma; Adult Mixed Glioma; Recurrent Adult Brain Tumor; Adult Anaplastic Oligoastrocytoma; Recurrent High Grade Glioma
An ice-templated, linearly aligned chitosan-alginate scaffold for neural tissue engineering.
Francis, Nicola L; Hunger, Philipp M; Donius, Amalie E; Riblett, Benjamin W; Zavaliangos, Antonios; Wegst, Ulrike G K; Wheatley, Margaret A
2013-12-01
Several strategies have been investigated to enhance axonal regeneration after spinal cord injury, however, the resulting growth can be random and disorganized. Bioengineered scaffolds provide a physical substrate for guidance of regenerating axons towards their targets, and can be produced by freeze casting. This technique involves the controlled directional solidification of an aqueous solution or suspension, resulting in a linearly aligned porous structure caused by ice templating. In this study, freeze casting was used to fabricate porous chitosan-alginate (C/A) scaffolds with longitudinally oriented channels. Chick dorsal root ganglia explants adhered to and extended neurites through the scaffold in parallel alignment with the channel direction. Surface adsorption of a polycation and laminin promoted significantly longer neurite growth than the uncoated scaffold (poly-L-ornithine + Laminin = 793.2 ± 187.2 μm; poly-L-lysine + Laminin = 768.7 ± 241.2 μm; uncoated scaffold = 22.52 ± 50.14 μm) (P < 0.001). The elastic modulus of the hydrated scaffold was determined to be 5.08 ± 0.61 kPa, comparable to reported spinal cord values. The present data suggested that this C/A scaffold is a promising candidate for use as a nerve guidance scaffold, because of its ability to support neuronal attachment and the linearly aligned growth of DRG neurites. Copyright © 2013 Wiley Periodicals, Inc., a Wiley Company.
Oparaji, Uchenna; Sheu, Rong-Jiun; Bankhead, Mark; Austin, Jonathan; Patelli, Edoardo
2017-12-01
Artificial Neural Networks (ANNs) are commonly used in place of expensive models to reduce the computational burden required for uncertainty quantification, reliability and sensitivity analyses. ANN with selected architecture is trained with the back-propagation algorithm from few data representatives of the input/output relationship of the underlying model of interest. However, different performing ANNs might be obtained with the same training data as a result of the random initialization of the weight parameters in each of the network, leading to an uncertainty in selecting the best performing ANN. On the other hand, using cross-validation to select the best performing ANN based on the ANN with the highest R 2 value can lead to biassing in the prediction. This is as a result of the fact that the use of R 2 cannot determine if the prediction made by ANN is biased. Additionally, R 2 does not indicate if a model is adequate, as it is possible to have a low R 2 for a good model and a high R 2 for a bad model. Hence, in this paper, we propose an approach to improve the robustness of a prediction made by ANN. The approach is based on a systematic combination of identical trained ANNs, by coupling the Bayesian framework and model averaging. Additionally, the uncertainties of the robust prediction derived from the approach are quantified in terms of confidence intervals. To demonstrate the applicability of the proposed approach, two synthetic numerical examples are presented. Finally, the proposed approach is used to perform a reliability and sensitivity analyses on a process simulation model of a UK nuclear effluent treatment plant developed by National Nuclear Laboratory (NNL) and treated in this study as a black-box employing a set of training data as a test case. This model has been extensively validated against plant and experimental data and used to support the UK effluent discharge strategy. Copyright © 2017 Elsevier Ltd. All rights reserved.
Suemitsu, Yoshikazu; Nara, Shigetoshi
2004-09-01
Chaotic dynamics introduced into a neural network model is applied to solving two-dimensional mazes, which are ill-posed problems. A moving object moves from the position at t to t + 1 by simply defined motion function calculated from firing patterns of the neural network model at each time step t. We have embedded several prototype attractors that correspond to the simple motion of the object orienting toward several directions in two-dimensional space in our neural network model. Introducing chaotic dynamics into the network gives outputs sampled from intermediate state points between embedded attractors in a state space, and these dynamics enable the object to move in various directions. System parameter switching between a chaotic and an attractor regime in the state space of the neural network enables the object to move to a set target in a two-dimensional maze. Results of computer simulations show that the success rate for this method over 300 trials is higher than that of random walk. To investigate why the proposed method gives better performance, we calculate and discuss statistical data with respect to dynamical structure.
Czech Academy of Sciences Publication Activity Database
Doubravová, Jana; Wiszniowski, J.; Horálek, Josef
2016-01-01
Roč. 93, August (2016), s. 138-149 ISSN 0098-3004 R&D Projects: GA ČR GAP210/12/2336; GA MŠk LM2010008 Institutional support: RVO:67985530 Keywords : event detection * artificial neural network * West Bohemia/Vogtland Subject RIV: DC - Siesmology, Volcanology, Earth Structure Impact factor: 2.533, year: 2016
Cui, Yiqian; Shi, Junyou; Wang, Zili
2017-11-01
Built-in tests (BITs) are widely used in mechanical systems to perform state identification, whereas the BIT false and missed alarms cause trouble to the operators or beneficiaries to make correct judgments. Artificial neural networks (ANN) are previously used for false and missed alarms identification, which has the features such as self-organizing and self-study. However, these ANN models generally do not incorporate the temporal effect of the bottom-level threshold comparison outputs and the historical temporal features are not fully considered. To improve the situation, this paper proposes a new integrated BIT design methodology by incorporating a novel type of dynamic neural networks (DNN) model. The new DNN model is termed as Forward IIR & Recurrent FIR DNN (FIRF-DNN), where its component neurons, network structures, and input/output relationships are discussed. The condition monitoring false and missed alarms reduction implementation scheme based on FIRF-DNN model is also illustrated, which is composed of three stages including model training, false and missed alarms detection, and false and missed alarms suppression. Finally, the proposed methodology is demonstrated in the application study and the experimental results are analyzed.
Directory of Open Access Journals (Sweden)
vahid Rezaverdinejad
2017-01-01
important models to estimate ETc in greenhouse. The inputs of these models are net radiation, temperature, day after planting and air vapour pressure deficit (or relative humidity. Materials and Methods: In this study, daily ETc of reference crop, greenhouse tomato and cucumber crops were measured using lysimeter method in Urmia region. Several linear, nonlinear regressions and artificial neural networks were considered for ETc modelling in greenhouse. For this purpose, the effective meteorological parameters on ETc process includes: air temperature (T, air humidity (RH, air pressure (P, air vapour pressure deficit (VPD, day after planting (N and greenhouse net radiation (SR were considered and measured. According to the goodness of fit, different models of artificial neural networks and regression were compared and evaluated. Furthermore, based on partial derivatives of regression models, sensitivity analysis was conducted. The accuracy and performance of the employed models was judged by ten statistical indices namely root mean square error (RMSE, normalized root mean square error (NRMSE and coefficient of determination (R2. Results and Discussion: Based on the results, the most accurate regression model to reference ETc prediction was obtained three variables exponential function of VPD, RH and SR with RMSE=0.378 mm day-1. The RMSE of optimal artificial neural network to reference ET prediction for train and test data sets were obtained 0.089 and 0.365 mm day-1, respectively. The performance of logarithmic and exponential functions to prediction of cucumber ETc were proper, with high dependent variables especially, and the most accurate regression model to cucumber ET prediction was obtained for exponential function of five variables: VPD, N, T, RH and SR with RMSE=0.353 mm day-1. In addition, for tomato ET prediction, the most accurate regression model was obtained for exponential function of four variables: VPD, N, RH and SR with RMSE= 0.329 mm day-1. The best
Ardell, Jeffrey L.; Shivkumar, Kalyanam; Armour, J. Andrew
2017-01-01
The cardiac nervous system continuously controls cardiac function whether or not pathology is present. While myocardial infarction typically has a major and catastrophic impact, population studies have shown that longer-term risk for recurrent myocardial infarction and the related potential for sudden cardiac death depends mainly upon standard atherosclerotic variables and autonomic nervous system maladaptations. Investigative neurocardiology has demonstrated that autonomic control of cardiac function includes local circuit neurons for networked control within the peripheral nervous system. The structural and adaptive characteristics of such networked interactions define the dynamics and a new normal for cardiac control that results in the aftermath of recurrent myocardial infarction and/or unstable angina that may or may not precipitate autonomic derangement. These features are explored here via a mathematical model of cardiac regulation. A main observation is that the control environment during pathology is an extrapolation to a setting outside prior experience. Although global bounds guarantee stability, the resulting closed-loop dynamics exhibited while the network adapts during pathology are aptly described as ‘free-floating’ in order to emphasize their dependence upon details of the network structure. The totality of the results provide a mechanistic reasoning that validates the clinical practice of reducing sympathetic efferent neuronal tone while aggressively targeting autonomic derangement in the treatment of ischemic heart disease. PMID:28692680
Sengupta, Rakesh; Surampudi, Bapi Raju; Melcher, David
2014-09-25
It has been proposed that the ability of humans to quickly perceive numerosity involves a visual sense of number. Different paradigms of enumeration and numerosity comparison have produced a gamut of behavioral and neuroimaging data, but there has been no unified conceptual framework that can explain results across the entire range of numerosity. The current work tries to address the ongoing debate concerning whether the same mechanism operates for enumeration of small and large numbers, through a computational approach. We describe the workings of a single-layered, fully connected network characterized by self-excitation and recurrent inhibition that operates at both subitizing and estimation ranges. We show that such a network can account for classic numerical cognition effects (the distance effect, Fechner׳s law, Weber fraction for numerosity comparison) through the network steady state activation response across different recurrent inhibition values. The model also accounts for fMRI data previously reported for different enumeration related tasks. The model also allows us to generate an estimate of the pattern of reaction times in enumeration tasks. Overall, these findings suggest that a single network architecture can account for both small and large number processing. Copyright © 2014. Published by Elsevier B.V.
Kember, Guy; Ardell, Jeffrey L; Shivkumar, Kalyanam; Armour, J Andrew
2017-01-01
The cardiac nervous system continuously controls cardiac function whether or not pathology is present. While myocardial infarction typically has a major and catastrophic impact, population studies have shown that longer-term risk for recurrent myocardial infarction and the related potential for sudden cardiac death depends mainly upon standard atherosclerotic variables and autonomic nervous system maladaptations. Investigative neurocardiology has demonstrated that autonomic control of cardiac function includes local circuit neurons for networked control within the peripheral nervous system. The structural and adaptive characteristics of such networked interactions define the dynamics and a new normal for cardiac control that results in the aftermath of recurrent myocardial infarction and/or unstable angina that may or may not precipitate autonomic derangement. These features are explored here via a mathematical model of cardiac regulation. A main observation is that the control environment during pathology is an extrapolation to a setting outside prior experience. Although global bounds guarantee stability, the resulting closed-loop dynamics exhibited while the network adapts during pathology are aptly described as 'free-floating' in order to emphasize their dependence upon details of the network structure. The totality of the results provide a mechanistic reasoning that validates the clinical practice of reducing sympathetic efferent neuronal tone while aggressively targeting autonomic derangement in the treatment of ischemic heart disease.
Directory of Open Access Journals (Sweden)
Guy Kember
Full Text Available The cardiac nervous system continuously controls cardiac function whether or not pathology is present. While myocardial infarction typically has a major and catastrophic impact, population studies have shown that longer-term risk for recurrent myocardial infarction and the related potential for sudden cardiac death depends mainly upon standard atherosclerotic variables and autonomic nervous system maladaptations. Investigative neurocardiology has demonstrated that autonomic control of cardiac function includes local circuit neurons for networked control within the peripheral nervous system. The structural and adaptive characteristics of such networked interactions define the dynamics and a new normal for cardiac control that results in the aftermath of recurrent myocardial infarction and/or unstable angina that may or may not precipitate autonomic derangement. These features are explored here via a mathematical model of cardiac regulation. A main observation is that the control environment during pathology is an extrapolation to a setting outside prior experience. Although global bounds guarantee stability, the resulting closed-loop dynamics exhibited while the network adapts during pathology are aptly described as 'free-floating' in order to emphasize their dependence upon details of the network structure. The totality of the results provide a mechanistic reasoning that validates the clinical practice of reducing sympathetic efferent neuronal tone while aggressively targeting autonomic derangement in the treatment of ischemic heart disease.
Ennouri, Karim; Ben Ayed, Rayda; Triki, Mohamed Ali; Ottaviani, Ennio; Mazzarello, Maura; Hertelli, Fathi; Zouari, Nabil
2017-07-01
The aim of the present work was to develop a model that supplies accurate predictions of the yields of delta-endotoxins and proteases produced by B. thuringiensis var. kurstaki HD-1. Using available medium ingredients as variables, a mathematical method, based on Plackett-Burman design (PB), was employed to analyze and compare data generated by the Bootstrap method and processed by multiple linear regressions (MLR) and artificial neural networks (ANN) including multilayer perceptron (MLP) and radial basis function (RBF) models. The predictive ability of these models was evaluated by comparison of output data through the determination of coefficient (R 2 ) and mean square error (MSE) values. The results demonstrate that the prediction of the yields of delta-endotoxin and protease was more accurate by ANN technique (87 and 89% for delta-endotoxin and protease determination coefficients, respectively) when compared with MLR method (73.1 and 77.2% for delta-endotoxin and protease determination coefficients, respectively), suggesting that the proposed ANNs, especially MLP, is a suitable new approach for determining yields of bacterial products that allow us to make more appropriate predictions in a shorter time and with less engineering effort.
Wang, Shui-Hua; Phillips, Preetha; Sui, Yuxiu; Liu, Bin; Yang, Ming; Cheng, Hong
2018-03-26
Alzheimer's disease (AD) is a progressive brain disease. The goal of this study is to provide a new computer-vision based technique to detect it in an efficient way. The brain-imaging data of 98 AD patients and 98 healthy controls was collected using data augmentation method. Then, convolutional neural network (CNN) was used, CNN is the most successful tool in deep learning. An 8-layer CNN was created with optimal structure obtained by experiences. Three activation functions (AFs): sigmoid, rectified linear unit (ReLU), and leaky ReLU. The three pooling-functions were also tested: average pooling, max pooling, and stochastic pooling. The numerical experiments demonstrated that leaky ReLU and max pooling gave the greatest result in terms of performance. It achieved a sensitivity of 97.96%, a specificity of 97.35%, and an accuracy of 97.65%, respectively. In addition, the proposed approach was compared with eight state-of-the-art approaches. The method increased the classification accuracy by approximately 5% compared to state-of-the-art methods.
Tshisaphungo, Mpho; Habarulema, John Bosco; McKinnell, Lee-Anne
2018-06-01
In this paper, the modeling of the ionospheric foF 2 changes during geomagnetic storms by means of neural network (NN) and linear regression (LR) techniques is presented. The results will lead to a valuable tool to model the complex ionospheric changes during disturbed days in an operational space weather monitoring and forecasting environment. The storm-time foF 2 data during 1996-2014 from Grahamstown (33.3°S, 26.5°E), South Africa ionosonde station was used in modeling. In this paper, six storms were reserved to validate the models and hence not used in the modeling process. We found that the performance of both NN and LR models is comparable during selected storms which fell within the data period (1996-2014) used in modeling. However, when validated on storm periods beyond 1996-2014, the NN model gives a better performance (R = 0.62) compared to LR model (R = 0.56) for a storm that reached a minimum Dst index of -155 nT during 19-23 December 2015. We also found that both NN and LR models are capable of capturing the ionospheric foF 2 responses during two great geomagnetic storms (28 October-1 November 2003 and 6-12 November 2004) which have been demonstrated to be difficult storms to model in previous studies.
Sequential neural models with stochastic layers
DEFF Research Database (Denmark)
Fraccaro, Marco; Sønderby, Søren Kaae; Paquet, Ulrich
2016-01-01
How can we efficiently propagate uncertainty in a latent state representation with recurrent neural networks? This paper introduces stochastic recurrent neural networks which glue a deterministic recurrent neural network and a state space model together to form a stochastic and sequential neural...... generative model. The clear separation of deterministic and stochastic layers allows a structured variational inference network to track the factorization of the model's posterior distribution. By retaining both the nonlinear recursive structure of a recurrent neural network and averaging over...
Directory of Open Access Journals (Sweden)
J. B. Habarulema
2009-05-01
Full Text Available This paper attempts to describe the search for the parameter(s to represent solar wind effects in Global Positioning System total electron content (GPS TEC modelling using the technique of neural networks (NNs. A study is carried out by including solar wind velocity (V_{sw}, proton number density (N_{p} and the B_{z} component of the interplanetary magnetic field (IMF B_{z} obtained from the Advanced Composition Explorer (ACE satellite as separate inputs to the NN each along with day number of the year (DN, hour (HR, a 4-month running mean of the daily sunspot number (R4 and the running mean of the previous eight 3-hourly magnetic A index values (A8. Hourly GPS TEC values derived from a dual frequency receiver located at Sutherland (32.38° S, 20.81° E, South Africa for 8 years (2000–2007 have been used to train the Elman neural network (ENN and the result has been used to predict TEC variations for a GPS station located at Cape Town (33.95° S, 18.47° E. Quantitative results indicate that each of the parameters considered may have some degree of influence on GPS TEC at certain periods although a decrease in prediction accuracy is also observed for some parameters for different days and seasons. It is also evident that there is still a difficulty in predicting TEC values during disturbed conditions. The improvements and degradation in prediction accuracies are both close to the benchmark values which lends weight to the belief that diurnal, seasonal, solar and magnetic variabilities may be the major determinants of TEC variability.
Directory of Open Access Journals (Sweden)
Chien-Lin Huang
2015-11-01
Full Text Available This study applies Real-Time Recurrent Learning Neural Network (RTRLNN and Adaptive Network-based Fuzzy Inference System (ANFIS with novel heuristic techniques to develop an advanced prediction model of accumulated total inflow of a reservoir in order to solve the difficulties of future long lead-time highly varied uncertainty during typhoon attacks while using a real-time forecast. For promoting the temporal-spatial forecasted precision, the following original specialized heuristic inputs were coupled: observed-predicted inflow increase/decrease (OPIID rate, total precipitation, and duration from current time to the time of maximum precipitation and direct runoff ending (DRE. This study also investigated the temporal-spatial forecasted error feature to assess the feasibility of the developed models, and analyzed the output sensitivity of both single and combined heuristic inputs to determine whether the heuristic model is susceptible to the impact of future forecasted uncertainty/errors. Validation results showed that the long lead-time–predicted accuracy and stability of the RTRLNN-based accumulated total inflow model are better than that of the ANFIS-based model because of the real-time recurrent deterministic routing mechanism of RTRLNN. Simulations show that the RTRLNN-based model with coupled heuristic inputs (RTRLNN-CHI, average error percentage (AEP/average forecast lead-time (AFLT: 6.3%/49 h can achieve better prediction than the model with non-heuristic inputs (AEP of RTRLNN-NHI and ANFIS-NHI: 15.2%/31.8% because of the full consideration of real-time hydrological initial/boundary conditions. Besides, the RTRLNN-CHI model can promote the forecasted lead-time above 49 h with less than 10% of AEP which can overcome the previous forecasted limits of 6-h AFLT with above 20%–40% of AEP.
Neural Networks in Control Applications
DEFF Research Database (Denmark)
Sørensen, O.
are examined. The models are separated into three groups representing input/output descriptions as well as state space descriptions: - Models, where all in- and outputs are measurable (static networks). - Models, where some inputs are non-measurable (recurrent networks). - Models, where some in- and some...... outputs are non-measurable (recurrent networks with incomplete state information). The three groups are ordered in increasing complexity, and for each group it is shown how to solve the problems concerning training and application of the specific model type. Of particular interest are the model types...... Kalmann filter) representing state space description. The potentials of neural networks for control of non-linear processes are also examined, focusing on three different groups of control concepts, all considered as generalizations of known linear control concepts to handle also non-linear processes...
Matsubara, Takashi; Torikai, Hiroyuki
2016-04-01
Modeling and implementation approaches for the reproduction of input-output relationships in biological nervous tissues contribute to the development of engineering and clinical applications. However, because of high nonlinearity, the traditional modeling and implementation approaches encounter difficulties in terms of generalization ability (i.e., performance when reproducing an unknown data set) and computational resources (i.e., computation time and circuit elements). To overcome these difficulties, asynchronous cellular automaton-based neuron (ACAN) models, which are described as special kinds of cellular automata that can be implemented as small asynchronous sequential logic circuits have been proposed. This paper presents a novel type of such ACAN and a theoretical analysis of its excitability. This paper also presents a novel network of such neurons, which can mimic input-output relationships of biological and nonlinear ordinary differential equation model neural networks. Numerical analyses confirm that the presented network has a higher generalization ability than other major modeling and implementation approaches. In addition, Field-Programmable Gate Array-implementations confirm that the presented network requires lower computational resources.
International Nuclear Information System (INIS)
Wei, Xinyu; Wang, Pengfei; Zhao, Fuyu
2016-01-01
Highlights: • We establish a disperse dynamic model for AP1000 reactor core. • A digital PID control based on QDRNN is used to design a decoupling control system. • The decoupling performance is verified and discussed. • The decoupling control system is simulated under the load following operation. - Abstract: The control system of the AP1000 reactor core uses the mechanical shim (MSHIM) strategy, which includes a power control subsystem and an axial power distribution control subsystem. To address the strong coupling between the two subsystems, an interlock between the two subsystems is used, which can only alleviate but not eliminate the coupling. Therefore, sometimes the axial offset (AO) cannot be controlled tightly, and the flexibility of load-following operation is limited. Thus, the decoupling of the original AP1000 reactor core control system is the focus of this paper. First, a two-node disperse dynamic model is established for the AP1000 reactor core to use PID control. Then, a digital PID control system based on a quasi-diagonal recurrent neural network (QDRNN) is designed to decouple the original system. Finally, the decoupling of the control system is verified by the step signal and load-following condition. The results show that the designed control system can decouple the original system as expected and the AO can be controlled much more tightly. Moreover, the flexibility of the load following is increased.
Directory of Open Access Journals (Sweden)
Faa-Jeng Lin
2014-01-01
Full Text Available This study presents a new active and reactive power control scheme for a single-stage three-phase grid-connected photovoltaic (PV system during grid faults. The presented PV system utilizes a single-stage three-phase current-controlled voltage-source inverter to achieve the maximum power point tracking (MPPT control of the PV panel with the function of low voltage ride through (LVRT. Moreover, a formula based on positive sequence voltage for evaluating the percentage of voltage sag is derived to determine the ratio of the injected reactive current to satisfy the LVRT regulations. To reduce the risk of overcurrent during LVRT operation, a current limit is predefined for the injection of reactive current. Furthermore, the control of active and reactive power is designed using a two-dimensional recurrent fuzzy cerebellar model articulation neural network (2D-RFCMANN. In addition, the online learning laws of 2D-RFCMANN are derived according to gradient descent method with varied learning-rate coefficients for network parameters to assure the convergence of the tracking error. Finally, some experimental tests are realized to validate the effectiveness of the proposed control scheme.
Wielgosz, Maciej; Skoczeń, Andrzej
This paper focuses on an examination of an applicability of Recurrent Neural Network models for detecting anomalous behavior of the CERN superconducting magnets. In order to conduct the experiments, the authors designed and implemented an adaptive signal quantization algorithm and a custom GRU-based detector and developed a method for the detector parameters selection. Three different datasets were used for testing the detector. Two artificially generated datasets were used to assess the raw performance of the system whereas the 231 MB dataset composed of the signals acquired from HiLumi magnets was intended for real-life experiments and model training. Several different setups of the developed anomaly detection system were evaluated and compared with state-of-the-art OC-SVM reference model operating on the same data. The OC-SVM model was equipped with a rich set of feature extractors accounting for a range of the input signal properties. It was determined in the course of the experiments that the detector, a...
Energy Technology Data Exchange (ETDEWEB)
Wei, Xinyu, E-mail: xyuwei@mail.xjtu.edu.cn; Wang, Pengfei, E-mail: pengfeixiaoli@yahoo.cn; Zhao, Fuyu, E-mail: fuyuzhao_xj@163.com
2016-08-01
Highlights: • We establish a disperse dynamic model for AP1000 reactor core. • A digital PID control based on QDRNN is used to design a decoupling control system. • The decoupling performance is verified and discussed. • The decoupling control system is simulated under the load following operation. - Abstract: The control system of the AP1000 reactor core uses the mechanical shim (MSHIM) strategy, which includes a power control subsystem and an axial power distribution control subsystem. To address the strong coupling between the two subsystems, an interlock between the two subsystems is used, which can only alleviate but not eliminate the coupling. Therefore, sometimes the axial offset (AO) cannot be controlled tightly, and the flexibility of load-following operation is limited. Thus, the decoupling of the original AP1000 reactor core control system is the focus of this paper. First, a two-node disperse dynamic model is established for the AP1000 reactor core to use PID control. Then, a digital PID control system based on a quasi-diagonal recurrent neural network (QDRNN) is designed to decouple the original system. Finally, the decoupling of the control system is verified by the step signal and load-following condition. The results show that the designed control system can decouple the original system as expected and the AO can be controlled much more tightly. Moreover, the flexibility of the load following is increased.
Fairbank, Michael; Li, Shuhui; Fu, Xingang; Alonso, Eduardo; Wunsch, Donald
2014-01-01
We present a recurrent neural-network (RNN) controller designed to solve the tracking problem for control systems. We demonstrate that a major difficulty in training any RNN is the problem of exploding gradients, and we propose a solution to this in the case of tracking problems, by introducing a stabilization matrix and by using carefully constrained context units. This solution allows us to achieve consistently lower training errors, and hence allows us to more easily introduce adaptive capabilities. The resulting RNN is one that has been trained off-line to be rapidly adaptive to changing plant conditions and changing tracking targets. The case study we use is a renewable-energy generator application; that of producing an efficient controller for a three-phase grid-connected converter. The controller we produce can cope with the random variation of system parameters and fluctuating grid voltages. It produces tracking control with almost instantaneous response to changing reference states, and virtually zero oscillation. This compares very favorably to the classical proportional integrator (PI) controllers, which we show produce a much slower response and settling time. In addition, the RNN we propose exhibits better learning stability and convergence properties, and can exhibit faster adaptation, than has been achieved with adaptive critic designs. Copyright © 2013 Elsevier Ltd. All rights reserved.
Wang, Chi-Hsu; Chen, Chun-Yao; Hung, Kun-Neng
2015-06-01
In this paper, a new adaptive self-organizing map (SOM) with recurrent neural network (RNN) controller is proposed for task assignment and path evolution of missile defense system (MDS). We address the problem of N agents (defending missiles) and D targets (incoming missiles) in MDS. A new RNN controller is designed to force an agent (or defending missile) toward a target (or incoming missile), and a monitoring controller is also designed to reduce the error between RNN controller and ideal controller. A new SOM with RNN controller is then designed to dispatch agents to their corresponding targets by minimizing total damaging cost. This is actually an important application of the multiagent system. The SOM with RNN controller is the main controller. After task assignment, the weighting factors of our new SOM with RNN controller are activated to dispatch the agents toward their corresponding targets. Using the Lyapunov constraints, the weighting factors for the proposed SOM with RNN controller are updated to guarantee the stability of the path evolution (or planning) system. Excellent simulations are obtained using this new approach for MDS, which show that our RNN has the lowest average miss distance among the several techniques.
Kirschner, Andreas; Frishman, Dmitrij
2008-10-01
Prediction of beta-turns from amino acid sequences has long been recognized as an important problem in structural bioinformatics due to their frequent occurrence as well as their structural and functional significance. Because various structural features of proteins are intercorrelated, secondary structure information has been often employed as an additional input for machine learning algorithms while predicting beta-turns. Here we present a novel bidirectional Elman-type recurrent neural network with multiple output layers (MOLEBRNN) capable of predicting multiple mutually dependent structural motifs and demonstrate its efficiency in recognizing three aspects of protein structure: beta-turns, beta-turn types, and secondary structure. The advantage of our method compared to other predictors is that it does not require any external input except for sequence profiles because interdependencies between different structural features are taken into account implicitly during the learning process. In a sevenfold cross-validation experiment on a standard test dataset our method exhibits the total prediction accuracy of 77.9% and the Mathew's Correlation Coefficient of 0.45, the highest performance reported so far. It also outperforms other known methods in delineating individual turn types. We demonstrate how simultaneous prediction of multiple targets influences prediction performance on single targets. The MOLEBRNN presented here is a generic method applicable in a variety of research fields where multiple mutually depending target classes need to be predicted. http://webclu.bio.wzw.tum.de/predator-web/.
Lu, I-Cheng; Wu, Che-Wei; Chang, Pi-Ying; Chen, Hsiu-Ya; Tseng, Kuang-Yi; Randolph, Gregory W; Cheng, Kuang-I; Chiang, Feng-Yu
2016-04-01
The use of neuromuscular blocking agent may effect intraoperative neuromonitoring (IONM) during thyroid surgery. An enhanced neuromuscular-blockade (NMB) recovery protocol was investigated in a porcine model and subsequently clinically applied during human thyroid neural monitoring surgery. Prospective animal and retrospective clinical study. In the animal experiment, 12 piglets were injected with rocuronium 0.6 mg/kg and randomly allocated to receive normal saline, sugammadex 2 mg/kg, or sugammadex 4 mg/kg to compare the recovery of laryngeal electromyography (EMG). In a subsequent clinical application study, 50 patients who underwent thyroidectomy with IONM followed an enhanced NMB recovery protocol-rocuronium 0.6 mg/kg at anesthesia induction and sugammadex 2 mg/kg at the operation start. The train-of-four (TOF) ratio was used for continuous quantitative monitoring of neuromuscular transmission. In our porcine model, it took 49 ± 15, 13.2 ± 5.6, and 4.2 ± 1.5 minutes for the 80% recovery of laryngeal EMG after injection of saline, sugammadex 2 mg/kg, and sugammadex 4 mg/kg, respectively. In subsequent clinical human application, the TOF ratio recovered from 0 to >0.9 within 5 minutes after administration of sugammadex 2 mg/kg at the operation start. All patients had positive and high EMG amplitude at the early stage of the operation, and intubation was without difficulty in 96% of patients. Both porcine modeling and clinical human application demonstrated that sugammadex 2 mg/kg allows effective and rapid restoration of neuromuscular function suppressed by rocuronium. Implementation of this enhanced NMB recovery protocol assures optimal conditions for tracheal intubation as well as IONM in thyroid surgery. NA. © 2016 The American Laryngological, Rhinological and Otological Society, Inc.
Energy Technology Data Exchange (ETDEWEB)
Costa, Ederson D' Martin; Lemes, Nelson Henrique Teixeira, E-mail: nelson.lemes@unifal-mg.edu.br [Instituto de Ciencias Exatas, Universidade Federal de Alfenas, Alfenas, MG (Brazil); Santos, Marcelo Henrique dos [Instituto de Ciencias Farmaceuticas, Universidade Federal de Alfenas, Alfenas, MG (Brazil); Braga, Joao Pedro [Departamento de Quimica, Universidade Federal de Minas Gerais (UFMG), Belo Horizonte, MG (Brazil)
2012-07-01
This work propose a recursive neural network to solve inverse equilibrium problem. The acidity constants of 7-epiclusianone in ethanol-water binary mixtures were determined from multiwavelength spectrophotometric data. A linear relationship between acidity constants and the % w/v of ethanol in the solvent mixture was observed. The proposed method efficiency is compared with the Simplex method, commonly used in nonlinear optimization techniques. The neural network method is simple, numerically stable and has a broad range of applicability. (author)
Directory of Open Access Journals (Sweden)
Kyle A McQuisten
2009-10-01
Full Text Available Exogenous short interfering RNAs (siRNAs induce a gene knockdown effect in cells by interacting with naturally occurring RNA processing machinery. However not all siRNAs induce this effect equally. Several heterogeneous kinds of machine learning techniques and feature sets have been applied to modeling siRNAs and their abilities to induce knockdown. There is some growing agreement to which techniques produce maximally predictive models and yet there is little consensus for methods to compare among predictive models. Also, there are few comparative studies that address what the effect of choosing learning technique, feature set or cross validation approach has on finding and discriminating among predictive models.Three learning techniques were used to develop predictive models for effective siRNA sequences including Artificial Neural Networks (ANNs, General Linear Models (GLMs and Support Vector Machines (SVMs. Five feature mapping methods were also used to generate models of siRNA activities. The 2 factors of learning technique and feature mapping were evaluated by complete 3x5 factorial ANOVA. Overall, both learning techniques and feature mapping contributed significantly to the observed variance in predictive models, but to differing degrees for precision and accuracy as well as across different kinds and levels of model cross-validation.The methods presented here provide a robust statistical framework to compare among models developed under distinct learning techniques and feature sets for siRNAs. Further comparisons among current or future modeling approaches should apply these or other suitable statistically equivalent methods to critically evaluate the performance of proposed models. ANN and GLM techniques tend to be more sensitive to the inclusion of noisy features, but the SVM technique is more robust under large numbers of features for measures of model precision and accuracy. Features found to result in maximally predictive models are
Nie, Xiaobing; Zheng, Wei Xing; Cao, Jinde
2015-11-01
The problem of coexistence and dynamical behaviors of multiple equilibrium points is addressed for a class of memristive Cohen-Grossberg neural networks with non-monotonic piecewise linear activation functions and time-varying delays. By virtue of the fixed point theorem, nonsmooth analysis theory and other analytical tools, some sufficient conditions are established to guarantee that such n-dimensional memristive Cohen-Grossberg neural networks can have 5(n) equilibrium points, among which 3(n) equilibrium points are locally exponentially stable. It is shown that greater storage capacity can be achieved by neural networks with the non-monotonic activation functions introduced herein than the ones with Mexican-hat-type activation function. In addition, unlike most existing multistability results of neural networks with monotonic activation functions, those obtained 3(n) locally stable equilibrium points are located both in saturated regions and unsaturated regions. The theoretical findings are verified by an illustrative example with computer simulations. Copyright © 2015 Elsevier Ltd. All rights reserved.
Stamova, Ivanka; Stamov, Gani
2017-12-01
In this paper, we propose a fractional-order neural network system with time-varying delays and reaction-diffusion terms. We first develop a new Mittag-Leffler synchronization strategy for the controlled nodes via impulsive controllers. Using the fractional Lyapunov method sufficient conditions are given. We also study the global Mittag-Leffler synchronization of two identical fractional impulsive reaction-diffusion neural networks using linear controllers, which was an open problem even for integer-order models. Since the Mittag-Leffler stability notion is a generalization of the exponential stability concept for fractional-order systems, our results extend and improve the exponential impulsive control theory of neural network system with time-varying delays and reaction-diffusion terms to the fractional-order case. The fractional-order derivatives allow us to model the long-term memory in the neural networks, and thus the present research provides with a conceptually straightforward mathematical representation of rather complex processes. Illustrative examples are presented to show the validity of the obtained results. We show that by means of appropriate impulsive controllers we can realize the stability goal and to control the qualitative behavior of the states. An image encryption scheme is extended using fractional derivatives. Copyright © 2017 Elsevier Ltd. All rights reserved.
Artificial neural network cardiopulmonary modeling and diagnosis
Kangas, Lars J.; Keller, Paul E.
1997-01-01
The present invention is a method of diagnosing a cardiopulmonary condition in an individual by comparing data from a progressive multi-stage test for the individual to a non-linear multi-variate model, preferably a recurrent artificial neural network having sensor fusion. The present invention relies on a cardiovascular model developed from physiological measurements of an individual. Any differences between the modeled parameters and the parameters of an individual at a given time are used for diagnosis.
Fragkaki, A G; Farmaki, E; Thomaidis, N; Tsantili-Kakoulidou, A; Angelis, Y S; Koupparis, M; Georgakopoulos, C
2012-09-21
The comparison among different modelling techniques, such as multiple linear regression, partial least squares and artificial neural networks, has been performed in order to construct and evaluate models for prediction of gas chromatographic relative retention times of trimethylsilylated anabolic androgenic steroids. The performance of the quantitative structure-retention relationship study, using the multiple linear regression and partial least squares techniques, has been previously conducted. In the present study, artificial neural networks models were constructed and used for the prediction of relative retention times of anabolic androgenic steroids, while their efficiency is compared with that of the models derived from the multiple linear regression and partial least squares techniques. For overall ranking of the models, a novel procedure [Trends Anal. Chem. 29 (2010) 101-109] based on sum of ranking differences was applied, which permits the best model to be selected. The suggested models are considered useful for the estimation of relative retention times of designer steroids for which no analytical data are available. Copyright © 2012 Elsevier B.V. All rights reserved.
Shastri, Niket; Pathak, Kamlesh
2018-05-01
The water vapor content in atmosphere plays very important role in climate. In this paper the application of GPS signal in meteorology is discussed, which is useful technique that is used to estimate the perceptible water vapor of atmosphere. In this paper various algorithms like artificial neural network, support vector machine and multiple linear regression are use to predict perceptible water vapor. The comparative studies in terms of root mean square error and mean absolute errors are also carried out for all the algorithms.
Hardware implementation of stochastic spiking neural networks.
Rosselló, Josep L; Canals, Vincent; Morro, Antoni; Oliver, Antoni
2012-08-01
Spiking Neural Networks, the last generation of Artificial Neural Networks, are characterized by its bio-inspired nature and by a higher computational capacity with respect to other neural models. In real biological neurons, stochastic processes represent an important mechanism of neural behavior and are responsible of its special arithmetic capabilities. In this work we present a simple hardware implementation of spiking neurons that considers this probabilistic nature. The advantage of the proposed implementation is that it is fully digital and therefore can be massively implemented in Field Programmable Gate Arrays. The high computational capabilities of the proposed model are demonstrated by the study of both feed-forward and recurrent networks that are able to implement high-speed signal filtering and to solve complex systems of linear equations.
International Nuclear Information System (INIS)
Canizes, Bruno; Soares, João; Faria, Pedro; Vale, Zita
2013-01-01
Highlights: • Ancillary services market management. • Ancillary services requirements forecast based on Artificial Neural Network. • Ancillary services clearing mechanisms without complex bids and with complex bids. - Abstract: Ancillary services represent a good business opportunity that must be considered by market players. This paper presents a new methodology for ancillary services market dispatch. The method considers the bids submitted to the market and includes a market clearing mechanism based on deterministic optimization. An Artificial Neural Network is used for day-ahead prediction of Regulation Down, regulation-up, Spin Reserve and Non-Spin Reserve requirements. Two test cases based on California Independent System Operator data concerning dispatch of Regulation Down, Regulation Up, Spin Reserve and Non-Spin Reserve services are included in this paper to illustrate the application of the proposed method: (1) dispatch considering simple bids; (2) dispatch considering complex bids
Energy Technology Data Exchange (ETDEWEB)
Kim, Yong-il [CHA University, Department of Nuclear Medicine, CHA Bundang Medical Center, Seongnam (Korea, Republic of); Seoul National University Hospital, Department of Nuclear Medicine, Seoul (Korea, Republic of); Kim, Yong Joong [Veterans Health Service Medical Center, Seoul (Korea, Republic of); Paeng, Jin Chul; Cheon, Gi Jeong; Lee, Dong Soo [Seoul National University Hospital, Department of Nuclear Medicine, Seoul (Korea, Republic of); Chung, June-Key [Seoul National University Hospital, Department of Nuclear Medicine, Seoul (Korea, Republic of); Seoul National University, Cancer Research Institute, Seoul (Korea, Republic of); Kang, Keon Wook [Seoul National University Hospital, Department of Nuclear Medicine, Seoul (Korea, Republic of); Seoul National University, Cancer Research Institute, Seoul (Korea, Republic of); Seoul National University College of Medicine, Department of Biomedical Sciences, Seoul (Korea, Republic of); Seoul National University College of Medicine, Department of Nuclear Medicine, Seoul (Korea, Republic of)
2017-11-15
{sup 18}F-Fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) has been investigated as a method to predict pancreatic cancer recurrence after pancreatic surgery. We evaluated the recently introduced heterogeneity indices of {sup 18}F-FDG PET/CT used for predicting pancreatic cancer recurrence after surgery and compared them with current clinicopathologic and {sup 18}F-FDG PET/CT parameters. A total of 93 pancreatic ductal adenocarcinoma patients (M:F = 60:33, mean age = 64.2 ± 9.1 years) who underwent preoperative {sup 18}F-FDG PET/CT following pancreatic surgery were retrospectively enrolled. The standardized uptake values (SUVs) and tumor-to-background ratios (TBR) were measured on each {sup 18}F-FDG PET/CT, as metabolic parameters. Metabolic tumor volume (MTV) and total lesion glycolysis (TLG) were examined as volumetric parameters. The coefficient of variance (heterogeneity index-1; SUVmean divided by the standard deviation) and linear regression slopes (heterogeneity index-2) of the MTV, according to SUV thresholds of 2.0, 2.5 and 3.0, were evaluated as heterogeneity indices. Predictive values of clinicopathologic and {sup 18}F-FDG PET/CT parameters and heterogeneity indices were compared in terms of pancreatic cancer recurrence. Seventy patients (75.3%) showed recurrence after pancreatic cancer surgery (mean recurrence = 9.4 ± 8.4 months). Comparing the recurrence and no recurrence patients, all of the {sup 18}F-FDG PET/CT parameters and heterogeneity indices demonstrated significant differences. In univariate Cox-regression analyses, MTV (P = 0.013), TLG (P = 0.007), and heterogeneity index-2 (P = 0.027) were significant. Among the clinicopathologic parameters, CA19-9 (P = 0.025) and venous invasion (P = 0.002) were selected as significant parameters. In multivariate Cox-regression analyses, MTV (P = 0.005), TLG (P = 0.004), and heterogeneity index-2 (P = 0.016) with venous invasion (P < 0.001, 0.001, and 0
Directory of Open Access Journals (Sweden)
Giovanni Leopoldo Rozza
2015-09-01
Full Text Available With world becoming each day a global village, enterprises continuously seek to optimize their internal processes to hold or improve their competitiveness and make better use of natural resources. In this context, decision support tools are an underlying requirement. Such tools are helpful on predicting operational issues, avoiding cost risings, loss of productivity, work-related accident leaves or environmental disasters. This paper has its focus on the prediction of spent liquor caustic concentration of Bayer process for alumina production. Caustic concentration measuring is essential to keep it at expected levels, otherwise quality issues might arise. The organization requests caustic concentration by chemical analysis laboratory once a day, such information is not enough to issue preventive actions to handle process inefficiencies that will be known only after new measurement on the next day. Thereby, this paper proposes using Multiple Linear Regression and Artificial Neural Networks techniques a mathematical model to predict the spent liquor´s caustic concentration. Hence preventive actions will occur in real time. Such models were built using software tool for numerical computation (MATLAB and a statistical analysis software package (SPSS. The models output (predicted caustic concentration were compared with the real lab data. We found evidence suggesting superior results with use of Artificial Neural Networks over Multiple Linear Regression model. The results demonstrate that replacing laboratorial analysis by the forecasting model to support technical staff on decision making could be feasible.
Hapugoda, J. C.; Sooriyarachchi, M. R.
2017-09-01
Survival time of patients with a disease and the incidence of that particular disease (count) is frequently observed in medical studies with the data of a clustered nature. In many cases, though, the survival times and the count can be correlated in a way that, diseases that occur rarely could have shorter survival times or vice versa. Due to this fact, joint modelling of these two variables will provide interesting and certainly improved results than modelling these separately. Authors have previously proposed a methodology using Generalized Linear Mixed Models (GLMM) by joining the Discrete Time Hazard model with the Poisson Regression model to jointly model survival and count model. As Aritificial Neural Network (ANN) has become a most powerful computational tool to model complex non-linear systems, it was proposed to develop a new joint model of survival and count of Dengue patients of Sri Lanka by using that approach. Thus, the objective of this study is to develop a model using ANN approach and compare the results with the previously developed GLMM model. As the response variables are continuous in nature, Generalized Regression Neural Network (GRNN) approach was adopted to model the data. To compare the model fit, measures such as root mean square error (RMSE), absolute mean error (AME) and correlation coefficient (R) were used. The measures indicate the GRNN model fits the data better than the GLMM model.
Deep Learning Methods for Improved Decoding of Linear Codes
Nachmani, Eliya; Marciano, Elad; Lugosch, Loren; Gross, Warren J.; Burshtein, David; Be'ery, Yair
2018-02-01
The problem of low complexity, close to optimal, channel decoding of linear codes with short to moderate block length is considered. It is shown that deep learning methods can be used to improve a standard belief propagation decoder, despite the large example space. Similar improvements are obtained for the min-sum algorithm. It is also shown that tying the parameters of the decoders across iterations, so as to form a recurrent neural network architecture, can be implemented with comparable results. The advantage is that significantly less parameters are required. We also introduce a recurrent neural decoder architecture based on the method of successive relaxation. Improvements over standard belief propagation are also observed on sparser Tanner graph representations of the codes. Furthermore, we demonstrate that the neural belief propagation decoder can be used to improve the performance, or alternatively reduce the computational complexity, of a close to optimal decoder of short BCH codes.
Golmohammadi, Hassan
2009-11-30
A quantitative structure-property relationship (QSPR) study was performed to develop models those relate the structure of 141 organic compounds to their octanol-water partition coefficients (log P(o/w)). A genetic algorithm was applied as a variable selection tool. Modeling of log P(o/w) of these compounds as a function of theoretically derived descriptors was established by multiple linear regression (MLR), partial least squares (PLS), and artificial neural network (ANN). The best selected descriptors that appear in the models are: atomic charge weighted partial positively charged surface area (PPSA-3), fractional atomic charge weighted partial positive surface area (FPSA-3), minimum atomic partial charge (Qmin), molecular volume (MV), total dipole moment of molecule (mu), maximum antibonding contribution of a molecule orbital in the molecule (MAC), and maximum free valency of a C atom in the molecule (MFV). The result obtained showed the ability of developed artificial neural network to prediction of partition coefficients of organic compounds. Also, the results revealed the superiority of ANN over the MLR and PLS models. Copyright 2009 Wiley Periodicals, Inc.
Directory of Open Access Journals (Sweden)
Katherine Rotker
2016-01-01
Full Text Available Varicocele recurrence is one of the most common complications associated with varicocele repair. A systematic review was performed to evaluate varicocele recurrence rates, anatomic causes of recurrence, and methods of management of recurrent varicoceles. The PubMed database was evaluated using keywords "recurrent" and "varicocele" as well as MESH criteria "recurrent" and "varicocele." Articles were not included that were not in English, represented single case reports, focused solely on subclinical varicocele, or focused solely on a pediatric population (age <18. Rates of recurrence vary with the technique of varicocele repair from 0% to 35%. Anatomy of recurrence can be defined by venography. Management of varicocele recurrence can be surgical or via embolization.
Salgado, Iván; Mera-Hernández, Manuel; Chairez, Isaac
2017-11-01
This study addresses the problem of designing an output-based controller to stabilize multi-input multi-output (MIMO) systems in the presence of parametric disturbances as well as uncertainties in the state model and output noise measurements. The controller design includes a linear state transformation which separates uncertainties matched to the control input and the unmatched ones. A differential neural network (DNN) observer produces a nonlinear approximation of the matched perturbation and the unknown states simultaneously in the transformed coordinates. This study proposes the use of the Attractive Ellipsoid Method (AEM) to optimize the gains of the controller and the gain observer in the DNN structure. As a consequence, the obtained control input minimizes the convergence zone for the estimation error. Moreover, the control design uses the estimated disturbance provided by the DNN to obtain a better performance in the stabilization task in comparison with a quasi-minimal output feedback controller based on a Luenberger observer and a sliding mode controller. Numerical results pointed out the advantages obtained by the nonlinear control based on the DNN observer. The first example deals with the stabilization of an academic linear MIMO perturbed system and the second example stabilizes the trajectories of a DC-motor into a predefined operation point. Copyright © 2017 ISA. Published by Elsevier Ltd. All rights reserved.
Azadi, Sama; Karimi-Jashni, Ayoub
2016-02-01
Predicting the mass of solid waste generation plays an important role in integrated solid waste management plans. In this study, the performance of two predictive models, Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) was verified to predict mean Seasonal Municipal Solid Waste Generation (SMSWG) rate. The accuracy of the proposed models is illustrated through a case study of 20 cities located in Fars Province, Iran. Four performance measures, MAE, MAPE, RMSE and R were used to evaluate the performance of these models. The MLR, as a conventional model, showed poor prediction performance. On the other hand, the results indicated that the ANN model, as a non-linear model, has a higher predictive accuracy when it comes to prediction of the mean SMSWG rate. As a result, in order to develop a more cost-effective strategy for waste management in the future, the ANN model could be used to predict the mean SMSWG rate. Copyright © 2015 Elsevier Ltd. All rights reserved.
Bonelli, Maria Grazia; Ferrini, Mauro; Manni, Andrea
2016-12-01
The assessment of metals and organic micropollutants contamination in agricultural soils is a difficult challenge due to the extensive area used to collect and analyze a very large number of samples. With Dioxins and dioxin-like PCBs measurement methods and subsequent the treatment of data, the European Community advises the develop low-cost and fast methods allowing routing analysis of a great number of samples, providing rapid measurement of these compounds in the environment, feeds and food. The aim of the present work has been to find a method suitable to describe the relations occurring between organic and inorganic contaminants and use the value of the latter in order to forecast the former. In practice, the use of a metal portable soil analyzer coupled with an efficient statistical procedure enables the required objective to be achieved. Compared to Multiple Linear Regression, the Artificial Neural Networks technique has shown to be an excellent forecasting method, though there is no linear correlation between the variables to be analyzed.
International Nuclear Information System (INIS)
Schierle, C.; Otto, M.
1992-01-01
A two layer perceptron with backpropagation of error is used for quantitative analysis in ICP-AES. The network was trained by emission spectra of two interfering lines of Cd and As and the concentrations of both elements were subsequently estimated from mixture spectra. The spectra of the Cd and As lines were also used to perform multiple linear regression (MLR) via the calculation of the pseudoinverse S + of the sensitivity matrix S. In the present paper it is shown that there exist close relations between the operation of the perceptron and the MLR procedure. These are most clearly apparent in the correlation between the weights of the backpropagation network and the elements of the pseudoinverse. Using MLR, the confidence intervals over the predictions are exploited to correct for the optical device of the wavelength shift. (orig.)
Kavuncuoglu, Hatice; Kavuncuoglu, Erhan; Karatas, Seyda Merve; Benli, Büsra; Sagdic, Osman; Yalcin, Hasan
2018-04-09
The mathematical model was established to determine the diameter of inhibition zone of the walnut extract on the twelve bacterial species. Type of extraction, concentration, and pathogens were taken as input variables. Two models were used with the aim of designing this system. One of them was developed with artificial neural networks (ANN), and the other was formed with multiple linear regression (MLR). Four common training algorithms were used. Levenberg-Marquardt (LM), Bayesian regulation (BR), scaled conjugate gradient (SCG) and resilient back propagation (RP) were investigated, and the algorithms were compared. Root mean squared error and correlation coefficient were evaluated as performance criteria. When these criteria were analyzed, ANN showed high prediction performance, while MLR showed low prediction performance. As a result, it is seen that when the different input values are provided to the system developed with ANN, the most accurate inhibition zone (IZ) estimates were obtained. The results of this study could offer new perspectives, particularly in the field of microbiology, because these could be applied to other type of extraction, concentrations, and pathogens, without resorting to experiments. Copyright © 2018 Elsevier B.V. All rights reserved.
Energy Technology Data Exchange (ETDEWEB)
Helama, S.; Holopainen, J.; Eronen, M. [Department of Geology, University of Helsinki, (Finland); Makarenko, N.G. [Russian Academy of Sciences, St. Petersburg (Russian Federation). Pulkovo Astronomical Observatory; Karimova, L.M.; Kruglun, O.A. [Institute of Mathematics, Almaty (Kazakhstan); Timonen, M. [Finnish Forest Research Institute, Rovaniemi Research Unit (Finland); Merilaeinen, J. [SAIMA Unit of the Savonlinna Department of Teacher Education, University of Joensuu (Finland)
2009-07-01
Tree-rings tell of past climates. To do so, tree-ring chronologies comprising numerous climate-sensitive living-tree and subfossil time-series need to be 'transferred' into palaeoclimate estimates using transfer functions. The purpose of this study is to compare different types of transfer functions, especially linear and nonlinear algorithms. Accordingly, multiple linear regression (MLR), linear scaling (LSC) and artificial neural networks (ANN, nonlinear algorithm) were compared. Transfer functions were built using a regional tree-ring chronology and instrumental temperature observations from Lapland (northern Finland and Sweden). In addition, conventional MLR was compared with a hybrid model whereby climate was reconstructed separately for short- and long-period timescales prior to combining the bands of timescales into a single hybrid model. The fidelity of the different reconstructions was validated against instrumental climate data. The reconstructions by MLR and ANN showed reliable reconstruction capabilities over the instrumental period (AD 1802-1998). LCS failed to reach reasonable verification statistics and did not qualify as a reliable reconstruction: this was due mainly to exaggeration of the low-frequency climatic variance. Over this instrumental period, the reconstructed low-frequency amplitudes of climate variability were rather similar by MLR and ANN. Notably greater differences between the models were found over the actual reconstruction period (AD 802-1801). A marked temperature decline, as reconstructed by MLR, from the Medieval Warm Period (AD 931-1180) to the Little Ice Age (AD 1601-1850), was evident in all the models. This decline was approx. 0.5 C as reconstructed by MLR. Different ANN based palaeotemperatures showed simultaneous cooling of 0.2 to 0.5 C, depending on algorithm. The hybrid MLR did not seem to provide further benefit above conventional MLR in our sample. The robustness of the conventional MLR over the calibration
Rosenberg, Jon; Galen, Benjamin T
2017-07-01
Recurrent meningitis is a rare clinical scenario that can be self-limiting or life threatening depending on the underlying etiology. This review describes the causes, risk factors, treatment, and prognosis for recurrent meningitis. As a general overview of a broad topic, the aim of this review is to provide clinicians with a comprehensive differential diagnosis to aide in the evaluation and management of a patient with recurrent meningitis. New developments related to understanding the pathophysiology of recurrent meningitis are as scarce as studies evaluating the treatment and prevention of this rare disorder. A trial evaluating oral valacyclovir suppression after HSV-2 meningitis did not demonstrate a benefit in preventing recurrences. The data on prophylactic antibiotics after basilar skull fractures do not support their use. Intrathecal trastuzumab has shown promise in treating leptomeningeal carcinomatosis from HER-2 positive breast cancer. Monoclonal antibodies used to treat cancer and autoimmune diseases are new potential causes of drug-induced aseptic meningitis. Despite their potential for causing recurrent meningitis, the clinical entities reviewed herein are not frequently discussed together given that they are a heterogeneous collection of unrelated, rare diseases. Epidemiologic data on recurrent meningitis are lacking. The syndrome of recurrent benign lymphocytic meningitis described by Mollaret in 1944 was later found to be closely related to HSV-2 reactivation, but HSV-2 is by no means the only etiology of recurrent aseptic meningitis. While the mainstay of treatment for recurrent meningitis is supportive care, it is paramount to ensure that reversible and treatable causes have been addressed for further prevention.
Energy Technology Data Exchange (ETDEWEB)
Ritter, G.X.; Sussner, P. [Univ. of Florida, Gainesville, FL (United States)
1996-12-31
The theory of artificial neural networks has been successfully applied to a wide variety of pattern recognition problems. In this theory, the first step in computing the next state of a neuron or in performing the next layer neural network computation involves the linear operation of multiplying neural values by their synaptic strengths and adding the results. Thresholding usually follows the linear operation in order to provide for nonlinearity of the network. In this paper we introduce a novel class of neural networks, called morphological neural networks, in which the operations of multiplication and addition are replaced by addition and maximum (or minimum), respectively. By taking the maximum (or minimum) of sums instead of the sum of products, morphological network computation is nonlinear before thresholding. As a consequence, the properties of morphological neural networks are drastically different than those of traditional neural network models. In this paper we consider some of these differences and provide some particular examples of morphological neural network.
Equivalence of Equilibrium Propagation and Recurrent Backpropagation
Scellier, Benjamin; Bengio, Yoshua
2017-01-01
Recurrent Backpropagation and Equilibrium Propagation are algorithms for fixed point recurrent neural networks which differ in their second phase. In the first phase, both algorithms converge to a fixed point which corresponds to the configuration where the prediction is made. In the second phase, Recurrent Backpropagation computes error derivatives whereas Equilibrium Propagation relaxes to another nearby fixed point. In this work we establish a close connection between these two algorithms....
2016-02-11
experiments were then conducted on the same BABEL task. The acoustic models were trained on 46 hours of speech. Tan - dem and hybrid DNN systems were...interpolation gave a comparable WER score of 46.9%. A fur - ther linear interpolation using equation (11) between the back-off based interpolated LM and the
International Nuclear Information System (INIS)
Ang, M R C O; Gonzalez, R M; Castro, P P M
2014-01-01
Rainfall, one of the important elements of the hydrologic cycle, is also the most difficult to model. Thus, accurate rainfall estimation is necessary especially in localized catchment areas where variability of rainfall is extremely high. Moreover, early warning of severe rainfall through timely and accurate estimation and forecasting could help prevent disasters from flooding. This paper presents the development of two rainfall estimation models that utilize a NARX-based neural network architecture namely: REIINN 1 and REIINN 2. These REIINN models, or Rainfall Estimation by Information Integration using Neural Networks, were trained using MTSAT cloud-top temperature (CTT) images and rainfall rates from the combined rain gauge and TMPA 3B40RT datasets. Model performance was assessed using two metrics – root mean square error (RMSE) and correlation coefficient (R). REIINN 1 yielded an RMSE of 8.1423 mm/3h and an overall R of 0.74652 while REIINN 2 yielded an RMSE of 5.2303 and an overall R of 0.90373. The results, especially that of REIINN 2, are very promising for satellite-based rainfall estimation in a catchment scale. It is believed that model performance and accuracy will greatly improve with a denser and more spatially distributed in-situ rainfall measurements to calibrate the model with. The models proved the viability of using remote sensing images, with their good spatial coverage, near real time availability, and relatively inexpensive to acquire, as an alternative source for rainfall estimation to complement existing ground-based measurements
Feature to prototype transition in neural networks
Krotov, Dmitry; Hopfield, John
Models of associative memory with higher order (higher than quadratic) interactions, and their relationship to neural networks used in deep learning are discussed. Associative memory is conventionally described by recurrent neural networks with dynamical convergence to stable points. Deep learning typically uses feedforward neural nets without dynamics. However, a simple duality relates these two different views when applied to problems of pattern classification. From the perspective of associative memory such models deserve attention because they make it possible to store a much larger number of memories, compared to the quadratic case. In the dual description, these models correspond to feedforward neural networks with one hidden layer and unusual activation functions transmitting the activities of the visible neurons to the hidden layer. These activation functions are rectified polynomials of a higher degree rather than the rectified linear functions used in deep learning. The network learns representations of the data in terms of features for rectified linear functions, but as the power in the activation function is increased there is a gradual shift to a prototype-based representation, the two extreme regimes of pattern recognition known in cognitive psychology. Simons Center for Systems Biology.
A Cheap Linear Attention Mechanism with Fast Lookups and Fixed-Size Representations
de Brébisson, Alexandre; Vincent, Pascal
2016-01-01
The softmax content-based attention mechanism has proven to be very beneficial in many applications of recurrent neural networks. Nevertheless it suffers from two major computational limitations. First, its computations for an attention lookup scale linearly in the size of the attended sequence. Second, it does not encode the sequence into a fixed-size representation but instead requires to memorize all the hidden states. These two limitations restrict the use of the softmax attention mechani...
Directory of Open Access Journals (Sweden)
Zheng Lu
2017-06-01
Full Text Available A method using a nonlinear auto-regressive neural network with exogenous input (NARXnn to retrieve time series soil moisture (SM that is spatially and temporally continuous and high quality over the Heihe River Basin (HRB in China was investigated in this study. The input training data consisted of the X-band dual polarization brightness temperature (TB and the Ka-band V polarization TB from the Advanced Microwave Scanning Radiometer II (AMSR2, Global Land Satellite product (GLASS Leaf Area Index (LAI, precipitation from the Tropical Rainfall Measuring Mission (TRMM and the Global Precipitation Measurement (GPM, and a global 30 arc-second elevation (GTOPO-30. The output training data were generated from fused SM products of the Japan Aerospace Exploration Agency (JAXA and the Land Surface Parameter Model (LPRM. The reprocessed fused SM from two years (2013 and 2014 was inputted into the NARXnn for training; subsequently, SM during a third year (2015 was estimated. Direct and indirect validations were then performed during the period 2015 by comparing with in situ measurements, SM from JAXA, LPRM and the Global Land Data Assimilation System (GLDAS, as well as precipitation data from TRMM and GPM. The results showed that the SM predictions from NARXnn performed best, as indicated by their higher correlation coefficients (R ≥ 0.85 for the whole year of 2015, lower Bias values (absolute value of Bias ≤ 0.02 and root mean square error values (RMSE ≤ 0.06, and their improved response to precipitation. This method is being used to produce the NARXnn SM product over the HRB in China.
Powell, Anna M; Nyirjesy, Paul
2014-10-01
Vulvovaginitis (VV) is one of the most commonly encountered problems by a gynecologist. Many women frequently self-treat with over-the-counter medications, and may present to their health-care provider after a treatment failure. Vulvovaginal candidiasis, bacterial vaginosis, and trichomoniasis may occur as discreet or recurrent episodes, and have been associated with significant treatment cost and morbidity. We present an update on diagnostic capabilities and treatment modalities that address recurrent and refractory episodes of VV. Copyright © 2014 Elsevier Ltd. All rights reserved.
Directory of Open Access Journals (Sweden)
Santana Isabel
2011-08-01
Full Text Available Abstract Background Dementia and cognitive impairment associated with aging are a major medical and social concern. Neuropsychological testing is a key element in the diagnostic procedures of Mild Cognitive Impairment (MCI, but has presently a limited value in the prediction of progression to dementia. We advance the hypothesis that newer statistical classification methods derived from data mining and machine learning methods like Neural Networks, Support Vector Machines and Random Forests can improve accuracy, sensitivity and specificity of predictions obtained from neuropsychological testing. Seven non parametric classifiers derived from data mining methods (Multilayer Perceptrons Neural Networks, Radial Basis Function Neural Networks, Support Vector Machines, CART, CHAID and QUEST Classification Trees and Random Forests were compared to three traditional classifiers (Linear Discriminant Analysis, Quadratic Discriminant Analysis and Logistic Regression in terms of overall classification accuracy, specificity, sensitivity, Area under the ROC curve and Press'Q. Model predictors were 10 neuropsychological tests currently used in the diagnosis of dementia. Statistical distributions of classification parameters obtained from a 5-fold cross-validation were compared using the Friedman's nonparametric test. Results Press' Q test showed that all classifiers performed better than chance alone (p Conclusions When taking into account sensitivity, specificity and overall classification accuracy Random Forests and Linear Discriminant analysis rank first among all the classifiers tested in prediction of dementia using several neuropsychological tests. These methods may be used to improve accuracy, sensitivity and specificity of Dementia predictions from neuropsychological testing.
Representation of neural networks as Lotka-Volterra systems
International Nuclear Information System (INIS)
Moreau, Yves; Vandewalle, Joos; Louies, Stephane; Brenig, Leon
1999-01-01
We study changes of coordinates that allow the representation of the ordinary differential equations describing continuous-time recurrent neural networks into differential equations describing predator-prey models--also called Lotka-Volterra systems. We transform the equations for the neural network first into quasi-monomial form, where we express the vector field of the dynamical system as a linear combination of products of powers of the variables. In practice, this transformation is possible only if the activation function is the hyperbolic tangent or the logistic sigmoied. From this quasi-monomial form, we can directly transform the system further into Lotka-Volterra equations. The resulting Lotka-Volterra system is of higher dimension than the original system, but the behavior of its first variables is equivalent to the behavior of the original neural network
Dynamic training algorithm for dynamic neural networks
International Nuclear Information System (INIS)
Tan, Y.; Van Cauwenberghe, A.; Liu, Z.
1996-01-01
The widely used backpropagation algorithm for training neural networks based on the gradient descent has a significant drawback of slow convergence. A Gauss-Newton method based recursive least squares (RLS) type algorithm with dynamic error backpropagation is presented to speed-up the learning procedure of neural networks with local recurrent terms. Finally, simulation examples concerning the applications of the RLS type algorithm to identification of nonlinear processes using a local recurrent neural network are also included in this paper
Neural Network Based Real-time Correction of Transducer Dynamic Errors
Roj, J.
2013-12-01
In order to carry out real-time dynamic error correction of transducers described by a linear differential equation, a novel recurrent neural network was developed. The network structure is based on solving this equation with respect to the input quantity when using the state variables. It is shown that such a real-time correction can be carried out using simple linear perceptrons. Due to the use of a neural technique, knowledge of the dynamic parameters of the transducer is not necessary. Theoretical considerations are illustrated by the results of simulation studies performed for the modeled second order transducer. The most important properties of the neural dynamic error correction, when emphasizing the fundamental advantages and disadvantages, are discussed.
Advances in Artificial Neural Networks – Methodological Development and Application
Directory of Open Access Journals (Sweden)
Yanbo Huang
2009-08-01
Full Text Available Artificial neural networks as a major soft-computing technology have been extensively studied and applied during the last three decades. Research on backpropagation training algorithms for multilayer perceptron networks has spurred development of other neural network training algorithms for other networks such as radial basis function, recurrent network, feedback network, and unsupervised Kohonen self-organizing network. These networks, especially the multilayer perceptron network with a backpropagation training algorithm, have gained recognition in research and applications in various scientific and engineering areas. In order to accelerate the training process and overcome data over-fitting, research has been conducted to improve the backpropagation algorithm. Further, artificial neural networks have been integrated with other advanced methods such as fuzzy logic and wavelet analysis, to enhance the ability of data interpretation and modeling and to avoid subjectivity in the operation of the training algorithm. In recent years, support vector machines have emerged as a set of high-performance supervised generalized linear classifiers in parallel with artificial neural networks. A review on development history of artificial neural networks is presented and the standard architectures and algorithms of artificial neural networks are described. Furthermore, advanced artificial neural networks will be introduced with support vector machines, and limitations of ANNs will be identified. The future of artificial neural network development in tandem with support vector machines will be discussed in conjunction with further applications to food science and engineering, soil and water relationship for crop management, and decision support for precision agriculture. Along with the network structures and training algorithms, the applications of artificial neural networks will be reviewed as well, especially in the fields of agricultural and biological
Recurrent Spatial Transformer Networks
DEFF Research Database (Denmark)
Sønderby, Søren Kaae; Sønderby, Casper Kaae; Maaløe, Lars
2015-01-01
We integrate the recently proposed spatial transformer network (SPN) [Jaderberg et. al 2015] into a recurrent neural network (RNN) to form an RNN-SPN model. We use the RNN-SPN to classify digits in cluttered MNIST sequences. The proposed model achieves a single digit error of 1.5% compared to 2.......9% for a convolutional networks and 2.0% for convolutional networks with SPN layers. The SPN outputs a zoomed, rotated and skewed version of the input image. We investigate different down-sampling factors (ratio of pixel in input and output) for the SPN and show that the RNN-SPN model is able to down-sample the input...
Recurrent coupling improves discrimination of temporal spike patterns
Directory of Open Access Journals (Sweden)
Chun-Wei eYuan
2012-05-01
Full Text Available Despite the ubiquitous presence of recurrent synaptic connections insensory neuronal systems, their general functional purpose is not wellunderstood. A recent conceptual advance has been achieved by theoriesof reservoir computing in which recurrent networks have been proposedto generate short-term memory as well as to improve neuronalrepresentation of the sensory input for subsequent computations.Here, we present a numerical study on the distinct effects ofinhibitory and excitatory recurrence in a canonical linearclassification task. It is found that both types of coupling improvethe ability to discriminate temporal spike patterns as compared to apurely feed-forward system, although in different ways. For a largeclass of inhibitory networks, the network's performance is optimal aslong as a fraction of roughly 50% of neurons per stimulus is activein the resulting population code. Thereby the contribution of inactiveneurons to the neural code is found to be even more informative thanthat of the active neurons, generating an inherent robustness ofclassification performance against temporal jitter of the inputspikes. Excitatory couplings are found to not only produce ashort-term memory buffer but also to improve linear separability ofthe population patterns by evoking more irregular firing as comparedto the purely inhibitory case. As the excitatory connectivity becomesmore sparse, firing becomes more variable and pattern separabilityimproves. We argue that the proposed paradigm is particularlywell-suited as a conceptual framework for processing of sensoryinformation in the auditory pathway.
Neural Networks for Optimal Control
DEFF Research Database (Denmark)
Sørensen, O.
1995-01-01
Two neural networks are trained to act as an observer and a controller, respectively, to control a non-linear, multi-variable process.......Two neural networks are trained to act as an observer and a controller, respectively, to control a non-linear, multi-variable process....
Automated Item Generation with Recurrent Neural Networks.
von Davier, Matthias
2018-03-12
Utilizing technology for automated item generation is not a new idea. However, test items used in commercial testing programs or in research are still predominantly written by humans, in most cases by content experts or professional item writers. Human experts are a limited resource and testing agencies incur high costs in the process of continuous renewal of item banks to sustain testing programs. Using algorithms instead holds the promise of providing unlimited resources for this crucial part of assessment development. The approach presented here deviates in several ways from previous attempts to solve this problem. In the past, automatic item generation relied either on generating clones of narrowly defined item types such as those found in language free intelligence tests (e.g., Raven's progressive matrices) or on an extensive analysis of task components and derivation of schemata to produce items with pre-specified variability that are hoped to have predictable levels of difficulty. It is somewhat unlikely that researchers utilizing these previous approaches would look at the proposed approach with favor; however, recent applications of machine learning show success in solving tasks that seemed impossible for machines not too long ago. The proposed approach uses deep learning to implement probabilistic language models, not unlike what Google brain and Amazon Alexa use for language processing and generation.
A study on switched linear system identification using game ...
African Journals Online (AJOL)
A study on switched linear system identification using game-theoretic strategies and neural computing. ... This study deals with application of game-theoretic strategies and neural computing to switched linear ... AJOL African Journals Online.
Moebes, T. A.
1994-01-01
To locate the accessory pathway(s) in preexicitation syndromes, epicardial and endocardial ventricular mapping is performed during anterograde ventricular activation via accessory pathway(s) from data originally received in signal form. As the number of channels increases, it is pertinent that more automated detection of coherent/incoherent signals is achieved as well as the prediction and prognosis of ventricular tachywardia (VT). Today's computers and computer program algorithms are not good in simple perceptual tasks such as recognizing a pattern or identifying a sound. This discrepancy, among other things, has been a major motivating factor in developing brain-based, massively parallel computing architectures. Neural net paradigms have proven to be effective at pattern recognition tasks. In signal processing, the picking of coherent/incoherent signals represents a pattern recognition task for computer systems. The picking of signals representing the onset ot VT also represents such a computer task. We attacked this problem by defining four signal attributes for each potential first maximal arrival peak and one signal attribute over the entire signal as input to a back propagation neural network. One attribute was the predicted amplitude value after the maximum amplitude over a data window. Then, by using a set of known (user selected) coherent/incoherent signals, and signals representing the onset of VT, we trained the back propagation network to recognize coherent/incoherent signals, and signals indicating the onset of VT. Since our output scheme involves a true or false decision, and since the output unit computes values between 0 and 1, we used a Fuzzy Arithmetic approach to classify data as coherent/incoherent signals. Furthermore, a Mean-Square Error Analysis was used to determine system stability. The neural net based picking coherent/incoherent signal system achieved high accuracy on picking coherent/incoherent signals on different patients. The system
Heffernan, Rhys; Yang, Yuedong; Paliwal, Kuldip; Zhou, Yaoqi
2017-09-15
The accuracy of predicting protein local and global structural properties such as secondary structure and solvent accessible surface area has been stagnant for many years because of the challenge of accounting for non-local interactions between amino acid residues that are close in three-dimensional structural space but far from each other in their sequence positions. All existing machine-learning techniques relied on a sliding window of 10-20 amino acid residues to capture some 'short to intermediate' non-local interactions. Here, we employed Long Short-Term Memory (LSTM) Bidirectional Recurrent Neural Networks (BRNNs) which are capable of capturing long range interactions without using a window. We showed that the application of LSTM-BRNN to the prediction of protein structural properties makes the most significant improvement for residues with the most long-range contacts (|i-j| >19) over a previous window-based, deep-learning method SPIDER2. Capturing long-range interactions allows the accuracy of three-state secondary structure prediction to reach 84% and the correlation coefficient between predicted and actual solvent accessible surface areas to reach 0.80, plus a reduction of 5%, 10%, 5% and 10% in the mean absolute error for backbone ϕ , ψ , θ and τ angles, respectively, from SPIDER2. More significantly, 27% of 182724 40-residue models directly constructed from predicted C α atom-based θ and τ have similar structures to their corresponding native structures (6Å RMSD or less), which is 3% better than models built by ϕ and ψ angles. We expect the method to be useful for assisting protein structure and function prediction. The method is available as a SPIDER3 server and standalone package at http://sparks-lab.org . yaoqi.zhou@griffith.edu.au or yuedong.yang@griffith.edu.au. Supplementary data are available at Bioinformatics online. © The Author (2017). Published by Oxford University Press. All rights reserved. For Permissions, please email
Simplified LQG Control with Neural Networks
DEFF Research Database (Denmark)
Sørensen, O.
1997-01-01
A new neural network application for non-linear state control is described. One neural network is modelled to form a Kalmann predictor and trained to act as an optimal state observer for a non-linear process. Another neural network is modelled to form a state controller and trained to produce...
Contemporary deep recurrent learning for recognition
Iftekharuddin, K. M.; Alam, M.; Vidyaratne, L.
2017-05-01
Large-scale feed-forward neural networks have seen intense application in many computer vision problems. However, these networks can get hefty and computationally intensive with increasing complexity of the task. Our work, for the first time in literature, introduces a Cellular Simultaneous Recurrent Network (CSRN) based hierarchical neural network for object detection. CSRN has shown to be more effective to solving complex tasks such as maze traversal and image processing when compared to generic feed forward networks. While deep neural networks (DNN) have exhibited excellent performance in object detection and recognition, such hierarchical structure has largely been absent in neural networks with recurrency. Further, our work introduces deep hierarchy in SRN for object recognition. The simultaneous recurrency results in an unfolding effect of the SRN through time, potentially enabling the design of an arbitrarily deep network. This paper shows experiments using face, facial expression and character recognition tasks using novel deep recurrent model and compares recognition performance with that of generic deep feed forward model. Finally, we demonstrate the flexibility of incorporating our proposed deep SRN based recognition framework in a humanoid robotic platform called NAO.
Brignole, Michele; Menozzi, Carlo; Moya, Angel; Andresen, Dietrich; Blanc, Jean Jacques; Krahn, Andrew D.; Wieling, Wouter; Beiras, Xulio; Deharo, Jean Claude; Russo, Vitantonio; Tomaino, Marco; Sutton, Richard; Tomaino, M.; Pescoller, F.; Donateo, P.; Oddone, D.; Russo, V.; Pierri, F.; Matino, M. G.; Vitale, E.; Massa, R.; Piccinni, G.; Melissano, D.; Menozzi, C.; Lolli, G.; Gulizia, M.; Francese, M.; Iorfida, M.; Golzio, P.; Gaggioli, G.; Laffi, M.; Rabjoli, F.; Cecchinato, C.; Ungar, A.; Rafanelli, M.; Chisciotti, V.; Morrione, A.; del Rosso, A.; Guernaccia, V.; Palella, M.; D'Agostino, C.; Campana, A.; Brigante, M.; Miracapillo, G.; Addonisio, L.; Proclemer, A.; Facchin, D.; Vado, A.; Knops, R. E.; Dekker, L. R. C.
2012-01-01
Background-The efficacy of cardiac pacing for prevention of syncopal recurrences in patients with neurally mediated syncope is controversial. We wanted to determine whether pacing therapy reduces syncopal recurrences in patients with severe asystolic neurally mediated syncope. Methods and
Recurrent competition explains temporal effects of attention in MSTd
Layton, Oliver W.; Browning, N. Andrew
2012-01-01
Navigation in a static environment along straight paths without eye movements produces radial optic flow fields. A singularity called the focus of expansion (FoE) specifies the direction of travel (heading) of the observer. Cells in primate dorsal medial superior temporal area (MSTd) respond to radial fields and are therefore thought to be heading-sensitive. Humans frequently shift their focus of attention while navigating, for example, depending on the favorable or threatening context of approaching independently moving objects. Recent neurophysiological studies show that the spatial tuning curves of primate MSTd neurons change based on the difference in visual angle between an attentional prime and the FoE. Moreover, the peak mean population activity in MSTd retreats linearly in time as the distance between the attentional prime and FoE increases. We present a dynamical neural circuit model that demonstrates the same linear temporal peak shift observed electrophysiologically. The model qualitatively matches the neuron tuning curves and population activation profiles. After model MT dynamically pools short-range motion, model MSTd incorporates recurrent competition between units tuned to different radial optic flow templates, and integrates attentional signals from model area frontal eye fields (FEF). In the model, population activity peaks occur when the recurrent competition is most active and uncertainty is greatest about the relative position of the FoE. The nature of attention, multiplicative or non-multiplicative, is largely irrelevant, so long as attention has a Gaussian-like profile. Using an appropriately tuned sigmoidal signal function to modulate recurrent feedback affords qualitative fits of deflections in the population activity that otherwise appear to be low-frequency noise. We predict that these deflections mark changes in the balance of attention between the priming and FoE locations. PMID:23060788
6th International Symposium on Recurrence Plots
Jr, Jr; Ioana, Cornel; Marwan, Norbert
2016-01-01
The chapters in this book originate from the research work and contributions presented at the Sixth International Symposium on Recurrence Plots held in Grenoble, France in June 2015. Scientists from numerous disciplines gathered to exchange knowledge on recent applications and developments in recurrence plots and recurrence quantification analysis. This meeting was remarkable because of the obvious expansion of recurrence strategies (theory) and applications (practice) into ever-broadening fields of science. It discusses real-world systems from various fields, including mathematics, strange attractors, applied physics, physiology, medicine, environmental and earth sciences, as well as psychology and linguistics. Even readers not actively researching any of these particular systems will benefit from discovering how other scientists are finding practical non-linear solutions to specific problems. The book is of interest to an interdisciplinary audience of recurrence plot users and researchers interested in time...
Computing Generalized Matrix Inverse on Spiking Neural Substrate
Shukla, Rohit; Khoram, Soroosh; Jorgensen, Erik; Li, Jing; Lipasti, Mikko; Wright, Stephen
2018-01-01
Emerging neural hardware substrates, such as IBM's TrueNorth Neurosynaptic System, can provide an appealing platform for deploying numerical algorithms. For example, a recurrent Hopfield neural network can be used to find the Moore-Penrose generalized inverse of a matrix, thus enabling a broad class of linear optimizations to be solved efficiently, at low energy cost. However, deploying numerical algorithms on hardware platforms that severely limit the range and precision of representation for numeric quantities can be quite challenging. This paper discusses these challenges and proposes a rigorous mathematical framework for reasoning about range and precision on such substrates. The paper derives techniques for normalizing inputs and properly quantizing synaptic weights originating from arbitrary systems of linear equations, so that solvers for those systems can be implemented in a provably correct manner on hardware-constrained neural substrates. The analytical model is empirically validated on the IBM TrueNorth platform, and results show that the guarantees provided by the framework for range and precision hold under experimental conditions. Experiments with optical flow demonstrate the energy benefits of deploying a reduced-precision and energy-efficient generalized matrix inverse engine on the IBM TrueNorth platform, reflecting 10× to 100× improvement over FPGA and ARM core baselines. PMID:29593483