Artificial neural network classification using a minimal training set - Comparison to conventional supervised classification

Science.gov (United States)

Hepner, George F.; Logan, Thomas; Ritter, Niles; Bryant, Nevin

1990-01-01

Recent research has shown an artificial neural network (ANN) to be capable of pattern recognition and the classification of image data. This paper examines the potential for the application of neural network computing to satellite image processing. A second objective is to provide a preliminary comparison and ANN classification. An artificial neural network can be trained to do land-cover classification of satellite imagery using selected sites representative of each class in a manner similar to conventional supervised classification. One of the major problems associated with recognition and classifications of pattern from remotely sensed data is the time and cost of developing a set of training sites. This reseach compares the use of an ANN back propagation classification procedure with a conventional supervised maximum likelihood classification procedure using a minimal training set. When using a minimal training set, the neural network is able to provide a land-cover classification superior to the classification derived from the conventional classification procedure. This research is the foundation for developing application parameters for further prototyping of software and hardware implementations for artificial neural networks in satellite image and geographic information processing.

C-RNN-GAN: Continuous recurrent neural networks with adversarial training

OpenAIRE

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.

Reward-based training of recurrent neural networks for cognitive and value-based tasks.

Science.gov (United States)

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.

Supervised learning in spiking neural networks with FORCE training.

Science.gov (United States)

Nicola, Wilten; Clopath, Claudia

2017-12-20

Populations of neurons display an extraordinary diversity in the behaviors they affect and display. Machine learning techniques have recently emerged that allow us to create networks of model neurons that display behaviors of similar complexity. Here we demonstrate the direct applicability of one such technique, the FORCE method, to spiking neural networks. We train these networks to mimic dynamical systems, classify inputs, and store discrete sequences that correspond to the notes of a song. Finally, we use FORCE training to create two biologically motivated model circuits. One is inspired by the zebra finch and successfully reproduces songbird singing. The second network is motivated by the hippocampus and is trained to store and replay a movie scene. FORCE trained networks reproduce behaviors comparable in complexity to their inspired circuits and yield information not easily obtainable with other techniques, such as behavioral responses to pharmacological manipulations and spike timing statistics.

Distributed computing methodology for training neural networks in an image-guided diagnostic application.

Science.gov (United States)

Plagianakos, V P; Magoulas, G D; Vrahatis, M N

2006-03-01

Distributed computing is a process through which a set of computers connected by a network is used collectively to solve a single problem. In this paper, we propose a distributed computing methodology for training neural networks for the detection of lesions in colonoscopy. Our approach is based on partitioning the training set across multiple processors using a parallel virtual machine. In this way, interconnected computers of varied architectures can be used for the distributed evaluation of the error function and gradient values, and, thus, training neural networks utilizing various learning methods. The proposed methodology has large granularity and low synchronization, and has been implemented and tested. Our results indicate that the parallel virtual machine implementation of the training algorithms developed leads to considerable speedup, especially when large network architectures and training sets are used.

Consistently Trained Artificial Neural Network for Automatic Ship Berthing Control

Directory of Open Access Journals (Sweden)

Y.A. Ahmed

2015-09-01

Full Text Available In this paper, consistently trained Artificial Neural Network controller for automatic ship berthing is discussed. Minimum time course changing manoeuvre is utilised to ensure such consistency and a new concept named ‘virtual window’ is introduced. Such consistent teaching data are then used to train two separate multi-layered feed forward neural networks for command rudder and propeller revolution output. After proper training, several known and unknown conditions are tested to judge the effectiveness of the proposed controller using Monte Carlo simulations. After getting acceptable percentages of success, the trained networks are implemented for the free running experiment system to judge the network’s real time response for Esso Osaka 3-m model ship. The network’s behaviour during such experiments is also investigated for possible effect of initial conditions as well as wind disturbances. Moreover, since the final goal point of the proposed controller is set at some distance from the actual pier to ensure safety, therefore a study on automatic tug assistance is also discussed for the final alignment of the ship with actual pier.

Statistical and optimization methods to expedite neural network training for transient identification

International Nuclear Information System (INIS)

Reifman, J.; Vitela, E.J.; Lee, J.C.

1993-01-01

Two complementary methods, statistical feature selection and nonlinear optimization through conjugate gradients, are used to expedite feedforward neural network training. Statistical feature selection techniques in the form of linear correlation coefficients and information-theoretic entropy are used to eliminate redundant and non-informative plant parameters to reduce the size of the network. The method of conjugate gradients is used to accelerate the network training convergence and to systematically calculate the Teaming and momentum constants at each iteration. The proposed techniques are compared with the backpropagation algorithm using the entire set of plant parameters in the training of neural networks to identify transients simulated with the Midland Nuclear Power Plant Unit 2 simulator. By using 25% of the plant parameters and the conjugate gradients, a 30-fold reduction in CPU time was obtained without degrading the diagnostic ability of the network

Gradual DropIn of Layers to Train Very Deep Neural Networks

OpenAIRE

Smith, Leslie N.; Hand, Emily M.; Doster, Timothy

2015-01-01

We introduce the concept of dynamically growing a neural network during training. In particular, an untrainable deep network starts as a trainable shallow network and newly added layers are slowly, organically added during training, thereby increasing the network's depth. This is accomplished by a new layer, which we call DropIn. The DropIn layer starts by passing the output from a previous layer (effectively skipping over the newly added layers), then increasingly including units from the ne...

Training strategy for convolutional neural networks in pedestrian gender classification

Science.gov (United States)

Ng, Choon-Boon; Tay, Yong-Haur; Goi, Bok-Min

2017-06-01

In this work, we studied a strategy for training a convolutional neural network in pedestrian gender classification with limited amount of labeled training data. Unsupervised learning by k-means clustering on pedestrian images was used to learn the filters to initialize the first layer of the network. As a form of pre-training, supervised learning for the related task of pedestrian classification was performed. Finally, the network was fine-tuned for gender classification. We found that this strategy improved the network's generalization ability in gender classification, achieving better test results when compared to random weights initialization and slightly more beneficial than merely initializing the first layer filters by unsupervised learning. This shows that unsupervised learning followed by pre-training with pedestrian images is an effective strategy to learn useful features for pedestrian gender classification.

The Analysis of User Behaviour of a Network Management Training Tool using a Neural Network

Directory of Open Access Journals (Sweden)

Helen Donelan

2005-10-01

Full Text Available A novel method for the analysis and interpretation of data that describes the interaction between trainee network managers and a network management training tool is presented. A simulation based approach is currently being used to train network managers, through the use of a simulated network. The motivation is to provide a tool for exposing trainees to a life like situation without disrupting a live network. The data logged by this system describes the detailed interaction between trainee network manager and simulated network. The work presented here provides an analysis of this interaction data that enables an assessment of the capabilities of the trainee network manager as well as an understanding of how the network management tasks are being approached. A neural network architecture is implemented in order to perform an exploratory data analysis of the interaction data. The neural network employs a novel form of continuous self-organisation to discover key features in the data and thus provide new insights into the learning and teaching strategies employed.

Pap-smear Classification Using Efficient Second Order Neural Network Training Algorithms

DEFF Research Database (Denmark)

Ampazis, Nikolaos; Dounias, George; Jantzen, Jan

2004-01-01

In this paper we make use of two highly efficient second order neural network training algorithms, namely the LMAM (Levenberg-Marquardt with Adaptive Momentum) and OLMAM (Optimized Levenberg-Marquardt with Adaptive Momentum), for the construction of an efficient pap-smear test classifier. The alg......In this paper we make use of two highly efficient second order neural network training algorithms, namely the LMAM (Levenberg-Marquardt with Adaptive Momentum) and OLMAM (Optimized Levenberg-Marquardt with Adaptive Momentum), for the construction of an efficient pap-smear test classifier....... The algorithms are methodologically similar, and are based on iterations of the form employed in the Levenberg-Marquardt (LM) method for non-linear least squares problems with the inclusion of an additional adaptive momentum term arising from the formulation of the training task as a constrained optimization...

PARTICLE SWARM OPTIMIZATION (PSO FOR TRAINING OPTIMIZATION ON CONVOLUTIONAL NEURAL NETWORK (CNN

Directory of Open Access Journals (Sweden)

Arie Rachmad Syulistyo

2016-02-01

Full Text Available Neural network attracts plenty of researchers lately. Substantial number of renowned universities have developed neural network for various both academically and industrially applications. Neural network shows considerable performance on various purposes. Nevertheless, for complex applications, neural network’s accuracy significantly deteriorates. To tackle the aforementioned drawback, lot of researches had been undertaken on the improvement of the standard neural network. One of the most promising modifications on standard neural network for complex applications is deep learning method. In this paper, we proposed the utilization of Particle Swarm Optimization (PSO in Convolutional Neural Networks (CNNs, which is one of the basic methods in deep learning. The use of PSO on the training process aims to optimize the results of the solution vectors on CNN in order to improve the recognition accuracy. The data used in this research is handwritten digit from MNIST. The experiments exhibited that the accuracy can be attained in 4 epoch is 95.08%. This result was better than the conventional CNN and DBN.  The execution time was also almost similar to the conventional CNN. Therefore, the proposed method was a promising method.

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...

Internal measuring models in trained neural networks for parameter estimation from images

NARCIS (Netherlands)

Feng, Tian-Jin; Feng, T.J.; Houkes, Z.; Korsten, Maarten J.; Spreeuwers, Lieuwe Jan

1992-01-01

The internal representations of 'learned' knowledge in neural networks are still poorly understood, even for backpropagation networks. The paper discusses a possible interpretation of learned knowledge of a network trained for parameter estimation from images. The outputs of the hidden layer are the

Bayesian model ensembling using meta-trained recurrent neural networks

NARCIS (Netherlands)

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

Planning Training Loads for the 400 M Hurdles in Three-Month Mesocycles using Artificial Neural Networks.

Science.gov (United States)

Przednowek, Krzysztof; Iskra, Janusz; Wiktorowicz, Krzysztof; Krzeszowski, Tomasz; Maszczyk, Adam

2017-12-01

This paper presents a novel approach to planning training loads in hurdling using artificial neural networks. The neural models performed the task of generating loads for athletes' training for the 400 meters hurdles. All the models were calculated based on the training data of 21 Polish National Team hurdlers, aged 22.25 ± 1.96, competing between 1989 and 2012. The analysis included 144 training plans that represented different stages in the annual training cycle. The main contribution of this paper is to develop neural models for planning training loads for the entire career of a typical hurdler. In the models, 29 variables were used, where four characterized the runner and 25 described the training process. Two artificial neural networks were used: a multi-layer perceptron and a network with radial basis functions. To assess the quality of the models, the leave-one-out cross-validation method was used in which the Normalized Root Mean Squared Error was calculated. The analysis shows that the method generating the smallest error was the radial basis function network with nine neurons in the hidden layer. Most of the calculated training loads demonstrated a non-linear relationship across the entire competitive period. The resulting model can be used as a tool to assist a coach in planning training loads during a selected training period.

Planning Training Loads for The 400 M Hurdles in Three-Month Mesocycles Using Artificial Neural Networks

Directory of Open Access Journals (Sweden)

Przednowek Krzysztof

2017-12-01

Full Text Available This paper presents a novel approach to planning training loads in hurdling using artificial neural networks. The neural models performed the task of generating loads for athletes’ training for the 400 meters hurdles. All the models were calculated based on the training data of 21 Polish National Team hurdlers, aged 22.25 ± 1.96, competing between 1989 and 2012. The analysis included 144 training plans that represented different stages in the annual training cycle. The main contribution of this paper is to develop neural models for planning training loads for the entire career of a typical hurdler. In the models, 29 variables were used, where four characterized the runner and 25 described the training process. Two artificial neural networks were used: a multi-layer perceptron and a network with radial basis functions. To assess the quality of the models, the leave-one-out cross-validation method was used in which the Normalized Root Mean Squared Error was calculated. The analysis shows that the method generating the smallest error was the radial basis function network with nine neurons in the hidden layer. Most of the calculated training loads demonstrated a non-linear relationship across the entire competitive period. The resulting model can be used as a tool to assist a coach in planning training loads during a selected training period.

Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices.

Science.gov (United States)

Gokmen, Tayfun; Onen, Murat; Haensch, Wilfried

2017-01-01

In a previous work we have detailed the requirements for obtaining maximal deep learning performance benefit by implementing fully connected deep neural networks (DNN) in the form of arrays of resistive devices. Here we extend the concept of Resistive Processing Unit (RPU) devices to convolutional neural networks (CNNs). We show how to map the convolutional layers to fully connected RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed by the analog nature of the computations performed on the arrays significantly affect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and that can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that a combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of the RPU approach to a large class of neural network architectures.

Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices

Science.gov (United States)

Gokmen, Tayfun; Onen, Murat; Haensch, Wilfried

2017-01-01

In a previous work we have detailed the requirements for obtaining maximal deep learning performance benefit by implementing fully connected deep neural networks (DNN) in the form of arrays of resistive devices. Here we extend the concept of Resistive Processing Unit (RPU) devices to convolutional neural networks (CNNs). We show how to map the convolutional layers to fully connected RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed by the analog nature of the computations performed on the arrays significantly affect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and that can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that a combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of the RPU approach to a large class of neural network architectures. PMID:29066942

Practical neural network recipies in C++

CERN Document Server

Masters

2014-01-01

This text serves as a cookbook for neural network solutions to practical problems using C++. It will enable those with moderate programming experience to select a neural network model appropriate to solving a particular problem, and to produce a working program implementing that network. The book provides guidance along the entire problem-solving path, including designing the training set, preprocessing variables, training and validating the network, and evaluating its performance. Though the book is not intended as a general course in neural networks, no background in neural works is assum

An Efficient Supervised Training Algorithm for Multilayer Spiking Neural Networks.

Science.gov (United States)

Xie, Xiurui; Qu, Hong; Liu, Guisong; Zhang, Malu; Kurths, Jürgen

2016-01-01

The spiking neural networks (SNNs) are the third generation of neural networks and perform remarkably well in cognitive tasks such as pattern recognition. The spike emitting and information processing mechanisms found in biological cognitive systems motivate the application of the hierarchical structure and temporal encoding mechanism in spiking neural networks, which have exhibited strong computational capability. However, the hierarchical structure and temporal encoding approach require neurons to process information serially in space and time respectively, which reduce the training efficiency significantly. For training the hierarchical SNNs, most existing methods are based on the traditional back-propagation algorithm, inheriting its drawbacks of the gradient diffusion and the sensitivity on parameters. To keep the powerful computation capability of the hierarchical structure and temporal encoding mechanism, but to overcome the low efficiency of the existing algorithms, a new training algorithm, the Normalized Spiking Error Back Propagation (NSEBP) is proposed in this paper. In the feedforward calculation, the output spike times are calculated by solving the quadratic function in the spike response model instead of detecting postsynaptic voltage states at all time points in traditional algorithms. Besides, in the feedback weight modification, the computational error is propagated to previous layers by the presynaptic spike jitter instead of the gradient decent rule, which realizes the layer-wised training. Furthermore, our algorithm investigates the mathematical relation between the weight variation and voltage error change, which makes the normalization in the weight modification applicable. Adopting these strategies, our algorithm outperforms the traditional SNN multi-layer algorithms in terms of learning efficiency and parameter sensitivity, that are also demonstrated by the comprehensive experimental results in this paper.

Modular representation of layered neural networks.

Science.gov (United States)

Watanabe, Chihiro; Hiramatsu, Kaoru; Kashino, Kunio

2018-01-01

Layered neural networks have greatly improved the performance of various applications including image processing, speech recognition, natural language processing, and bioinformatics. However, it is still difficult to discover or interpret knowledge from the inference provided by a layered neural network, since its internal representation has many nonlinear and complex parameters embedded in hierarchical layers. Therefore, it becomes important to establish a new methodology by which layered neural networks can be understood. In this paper, we propose a new method for extracting a global and simplified structure from a layered neural network. Based on network analysis, the proposed method detects communities or clusters of units with similar connection patterns. We show its effectiveness by applying it to three use cases. (1) Network decomposition: it can decompose a trained neural network into multiple small independent networks thus dividing the problem and reducing the computation time. (2) Training assessment: the appropriateness of a trained result with a given hyperparameter or randomly chosen initial parameters can be evaluated by using a modularity index. And (3) data analysis: in practical data it reveals the community structure in the input, hidden, and output layers, which serves as a clue for discovering knowledge from a trained neural network. Copyright © 2017 Elsevier Ltd. All rights reserved.

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....

Shakeout: A New Approach to Regularized Deep Neural Network Training.

Science.gov (United States)

Kang, Guoliang; Li, Jun; Tao, Dacheng

2018-05-01

Recent years have witnessed the success of deep neural networks in dealing with a plenty of practical problems. Dropout has played an essential role in many successful deep neural networks, by inducing regularization in the model training. In this paper, we present a new regularized training approach: Shakeout. Instead of randomly discarding units as Dropout does at the training stage, Shakeout randomly chooses to enhance or reverse each unit's contribution to the next layer. This minor modification of Dropout has the statistical trait: the regularizer induced by Shakeout adaptively combines , and regularization terms. Our classification experiments with representative deep architectures on image datasets MNIST, CIFAR-10 and ImageNet show that Shakeout deals with over-fitting effectively and outperforms Dropout. We empirically demonstrate that Shakeout leads to sparser weights under both unsupervised and supervised settings. Shakeout also leads to the grouping effect of the input units in a layer. Considering the weights in reflecting the importance of connections, Shakeout is superior to Dropout, which is valuable for the deep model compression. Moreover, we demonstrate that Shakeout can effectively reduce the instability of the training process of the deep architecture.

On Training Bi-directional Neural Network Language Model with Noise Contrastive Estimation

OpenAIRE

He, Tianxing; Zhang, Yu; Droppo, Jasha; Yu, Kai

2016-01-01

We propose to train bi-directional neural network language model(NNLM) with noise contrastive estimation(NCE). Experiments are conducted on a rescore task on the PTB data set. It is shown that NCE-trained bi-directional NNLM outperformed the one trained by conventional maximum likelihood training. But still(regretfully), it did not out-perform the baseline uni-directional NNLM.

Antenna analysis using neural networks

Science.gov (United States)

Smith, William T.

1992-01-01

Conventional computing schemes have long been used to analyze problems in electromagnetics (EM). The vast majority of EM applications require computationally intensive algorithms involving numerical integration and solutions to large systems of equations. The feasibility of using neural network computing algorithms for antenna analysis is investigated. The ultimate goal is to use a trained neural network algorithm to reduce the computational demands of existing reflector surface error compensation techniques. Neural networks are computational algorithms based on neurobiological systems. Neural nets consist of massively parallel interconnected nonlinear computational elements. They are often employed in pattern recognition and image processing problems. Recently, neural network analysis has been applied in the electromagnetics area for the design of frequency selective surfaces and beam forming networks. The backpropagation training algorithm was employed to simulate classical antenna array synthesis techniques. The Woodward-Lawson (W-L) and Dolph-Chebyshev (D-C) array pattern synthesis techniques were used to train the neural network. The inputs to the network were samples of the desired synthesis pattern. The outputs are the array element excitations required to synthesize the desired pattern. Once trained, the network is used to simulate the W-L or D-C techniques. Various sector patterns and cosecant-type patterns (27 total) generated using W-L synthesis were used to train the network. Desired pattern samples were then fed to the neural network. The outputs of the network were the simulated W-L excitations. A 20 element linear array was used. There were 41 input pattern samples with 40 output excitations (20 real parts, 20 imaginary). A comparison between the simulated and actual W-L techniques is shown for a triangular-shaped pattern. Dolph-Chebyshev is a different class of synthesis technique in that D-C is used for side lobe control as opposed to pattern

Distribution network fault section identification and fault location using artificial neural network

DEFF Research Database (Denmark)

Dashtdar, Masoud; Dashti, Rahman; Shaker, Hamid Reza

2018-01-01

In this paper, a method for fault location in power distribution network is presented. The proposed method uses artificial neural network. In order to train the neural network, a series of specific characteristic are extracted from the recorded fault signals in relay. These characteristics...... components of the sequences as well as three-phase signals could be obtained using statistics to extract the hidden features inside them and present them separately to train the neural network. Also, since the obtained inputs for the training of the neural network strongly depend on the fault angle, fault...... resistance, and fault location, the training data should be selected such that these differences are properly presented so that the neural network does not face any issues for identification. Therefore, selecting the signal processing function, data spectrum and subsequently, statistical parameters...

SuperNeurons: Dynamic GPU Memory Management for Training Deep Neural Networks

OpenAIRE

Wang, Linnan; Ye, Jinmian; Zhao, Yiyang; Wu, Wei; Li, Ang; Song, Shuaiwen Leon; Xu, Zenglin; Kraska, Tim

2018-01-01

Going deeper and wider in neural architectures improves the accuracy, while the limited GPU DRAM places an undesired restriction on the network design domain. Deep Learning (DL) practitioners either need change to less desired network architectures, or nontrivially dissect a network across multiGPUs. These distract DL practitioners from concentrating on their original machine learning tasks. We present SuperNeurons: a dynamic GPU memory scheduling runtime to enable the network training far be...

Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices

Directory of Open Access Journals (Sweden)

Tayfun Gokmen

2017-10-01

Full Text Available In a previous work we have detailed the requirements for obtaining maximal deep learning performance benefit by implementing fully connected deep neural networks (DNN in the form of arrays of resistive devices. Here we extend the concept of Resistive Processing Unit (RPU devices to convolutional neural networks (CNNs. We show how to map the convolutional layers to fully connected RPU arrays such that the parallelism of the hardware can be fully utilized in all three cycles of the backpropagation algorithm. We find that the noise and bound limitations imposed by the analog nature of the computations performed on the arrays significantly affect the training accuracy of the CNNs. Noise and bound management techniques are presented that mitigate these problems without introducing any additional complexity in the analog circuits and that can be addressed by the digital circuits. In addition, we discuss digitally programmable update management and device variability reduction techniques that can be used selectively for some of the layers in a CNN. We show that a combination of all those techniques enables a successful application of the RPU concept for training CNNs. The techniques discussed here are more general and can be applied beyond CNN architectures and therefore enables applicability of the RPU approach to a large class of neural network architectures.

Face recognition based on improved BP neural network

Directory of Open Access Journals (Sweden)

Yue Gaili

2017-01-01

Full Text Available In order to improve the recognition rate of face recognition, face recognition algorithm based on histogram equalization, PCA and BP neural network is proposed. First, the face image is preprocessed by histogram equalization. Then, the classical PCA algorithm is used to extract the features of the histogram equalization image, and extract the principal component of the image. And then train the BP neural network using the trained training samples. This improved BP neural network weight adjustment method is used to train the network because the conventional BP algorithm has the disadvantages of slow convergence, easy to fall into local minima and training process. Finally, the BP neural network with the test sample input is trained to classify and identify the face images, and the recognition rate is obtained. Through the use of ORL database face image simulation experiment, the analysis results show that the improved BP neural network face recognition method can effectively improve the recognition rate of face recognition.

Diagnostics of Nuclear Reactor Accidents Based on Particle Swarm Optimization Trained Neural Networks

International Nuclear Information System (INIS)

Abdel-Aal, M.M.Z.

2004-01-01

Automation in large, complex systems such as chemical plants, electrical power generation, aerospace and nuclear plants has been steadily increasing in the recent past. automated diagnosis and control forms a necessary part of these systems,this contains thousands of alarms processing in every component, subsystem and system. so the accurate and speed of diagnosis of faults is an important factors in operation and maintaining their health and continued operation and in reducing of repair and recovery time. using of artificial intelligence facilitates the alarm classifications and faults diagnosis to control any abnormal events during the operation cycle of the plant. thesis work uses the artificial neural network as a powerful classification tool. the work basically is has two components, the first is to effectively train the neural network using particle swarm optimization, which non-derivative based technique. to achieve proper training of the neural network to fault classification problem and comparing this technique to already existing techniques

Neural Networks in Control Applications

DEFF Research Database (Denmark)

Sørensen, O.

The intention of this report is to make a systematic examination of the possibilities of applying neural networks in those technical areas, which are familiar to a control engineer. In other words, the potential of neural networks in control applications is given higher priority than a detailed...... study of the networks themselves. With this end in view the following restrictions have been made: - Amongst numerous neural network structures, only the Multi Layer Perceptron (a feed-forward network) is applied. - Amongst numerous training algorithms, only four algorithms are examined, all...... in a recursive form (sample updating). The simplest is the Back Probagation Error Algorithm, and the most complex is the recursive Prediction Error Method using a Gauss-Newton search direction. - Over-fitting is often considered to be a serious problem when training neural networks. This problem is specifically...

Training Excitatory-Inhibitory Recurrent Neural Networks for Cognitive Tasks: A Simple and Flexible Framework.

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

Comparative Analysis of Neural Network Training Methods in Real-time Radiotherapy

Directory of Open Access Journals (Sweden)

Nouri S.

2017-03-01

Full Text Available Background: The motions of body and tumor in some regions such as chest during radiotherapy treatments are one of the major concerns protecting normal tissues against high doses. By using real-time radiotherapy technique, it is possible to increase the accuracy of delivered dose to the tumor region by means of tracing markers on the body of patients. Objective: This study evaluates the accuracy of some artificial intelligence methods including neural network and those of combination with genetic algorithm as well as particle swarm optimization (PSO estimating tumor positions in real-time radiotherapy. Method: One hundred recorded signals of three external markers were used as input data. The signals from 3 markers thorough 10 breathing cycles of a patient treated via a cyber-knife for a lung tumor were used as data input. Then, neural network method and its combination with genetic or PSO algorithms were applied determining the tumor locations using MATLAB© software program. Results: The accuracies were obtained 0.8%, 12% and 14% in neural network, genetic and particle swarm optimization algorithms, respectively. Conclusion: The internal target volume (ITV should be determined based on the applied neural network algorithm on training steps.

A modified backpropagation algorithm for training neural networks on data with error bars

International Nuclear Information System (INIS)

Gernoth, K.A.; Clark, J.W.

1994-08-01

A method is proposed for training multilayer feedforward neural networks on data contaminated with noise. Specifically, we consider the case that the artificial neural system is required to learn a physical mapping when the available values of the target variable are subject to experimental uncertainties, but are characterized by error bars. The proposed method, based on maximum likelihood criterion for parameter estimation, involves simple modifications of the on-line backpropagation learning algorithm. These include incorporation of the error-bar assignments in a pattern-specific learning rate, together with epochal updating of a new measure of model accuracy that replaces the usual mean-square error. The extended backpropagation algorithm is successfully tested on two problems relevant to the modelling of atomic-mass systematics by neural networks. Provided the underlying mapping is reasonably smooth, neural nets trained with the new procedure are able to learn the true function to a good approximation even in the presence of high levels of Gaussian noise. (author). 26 refs, 2 figs, 5 tabs

Internal-state analysis in layered artificial neural network trained to categorize lung sounds

NARCIS (Netherlands)

Oud, M

2002-01-01

In regular use of artificial neural networks, only input and output states of the network are known to the user. Weight and bias values can be extracted but are difficult to interpret. We analyzed internal states of networks trained to map asthmatic lung sound spectra onto lung function parameters.

Neural Networks for the Beginner.

Science.gov (United States)

Snyder, Robin M.

Motivated by the brain, neural networks are a right-brained approach to artificial intelligence that is used to recognize patterns based on previous training. In practice, one would not program an expert system to recognize a pattern and one would not train a neural network to make decisions from rules; but one could combine the best features of…

Neutron spectrometry and dosimetry by means of Bonner spheres system and artificial neural networks applying robust design of artificial neural networks

Energy Technology Data Exchange (ETDEWEB)

Martinez B, M.R.; Ortiz R, J.M.; Vega C, H.R. [UAZ, Av. Ramon Lopez Velarde No. 801, 98000 Zacatecas (Mexico)

2006-07-01

An Artificial Neural Network has been designed, trained and tested to unfold neutron spectra and simultaneously to calculate equivalent doses. A set of 187 neutron spectra compiled by the International Atomic Energy Agency and 13 equivalent doses were used in the artificial neural network designed, trained and tested. In order to design the neural network was used the robust design of artificial neural networks methodology, which assures that the quality of the neural networks takes into account from the design stage. Unless previous works, here, for first time a group of neural networks were designed and trained to unfold 187 neutron spectra and at the same time to calculate 13 equivalent doses, starting from the count rates coming from the Bonner spheres system by using a systematic and experimental strategy. (Author)

Neutron spectrometry and dosimetry by means of Bonner spheres system and artificial neural networks applying robust design of artificial neural networks

International Nuclear Information System (INIS)

Martinez B, M.R.; Ortiz R, J.M.; Vega C, H.R.

2006-01-01

An Artificial Neural Network has been designed, trained and tested to unfold neutron spectra and simultaneously to calculate equivalent doses. A set of 187 neutron spectra compiled by the International Atomic Energy Agency and 13 equivalent doses were used in the artificial neural network designed, trained and tested. In order to design the neural network was used the robust design of artificial neural networks methodology, which assures that the quality of the neural networks takes into account from the design stage. Unless previous works, here, for first time a group of neural networks were designed and trained to unfold 187 neutron spectra and at the same time to calculate 13 equivalent doses, starting from the count rates coming from the Bonner spheres system by using a systematic and experimental strategy. (Author)

Additive Feed Forward Control with Neural Networks

DEFF Research Database (Denmark)

Sørensen, O.

1999-01-01

This paper demonstrates a method to control a non-linear, multivariable, noisy process using trained neural networks. The basis for the method is a trained neural network controller acting as the inverse process model. A training method for obtaining such an inverse process model is applied....... A suitable 'shaped' (low-pass filtered) reference is used to overcome problems with excessive control action when using a controller acting as the inverse process model. The control concept is Additive Feed Forward Control, where the trained neural network controller, acting as the inverse process model......, is placed in a supplementary pure feed-forward path to an existing feedback controller. This concept benefits from the fact, that an existing, traditional designed, feedback controller can be retained without any modifications, and after training the connection of the neural network feed-forward controller...

Parameterization Of Solar Radiation Using Neural Network

International Nuclear Information System (INIS)

Jiya, J. D.; Alfa, B.

2002-01-01

This paper presents a neural network technique for parameterization of global solar radiation. The available data from twenty-one stations is used for training the neural network and the data from other ten stations is used to validate the neural model. The neural network utilizes latitude, longitude, altitude, sunshine duration and period number to parameterize solar radiation values. The testing data was not used in the training to demonstrate the performance of the neural network in unknown stations to parameterize solar radiation. The results indicate a good agreement between the parameterized solar radiation values and actual measured values

Neutron spectrometry with artificial neural networks

International Nuclear Information System (INIS)

Vega C, H.R.; Hernandez D, V.M.; Manzanares A, E.; Rodriguez, J.M.; Mercado S, G.A.; Iniguez de la Torre Bayo, M.P.; Barquero, R.; Arteaga A, T.

2005-01-01

An artificial neural network has been designed to obtain the neutron spectra from the Bonner spheres spectrometer's count rates. The neural network was trained using 129 neutron spectra. These include isotopic neutron sources; reference and operational spectra from accelerators and nuclear reactors, spectra from mathematical functions as well as few energy groups and monoenergetic spectra. The spectra were transformed from lethargy to energy distribution and were re-bin ned to 31 energy groups using the MCNP 4C code. Re-binned spectra and UTA4 response matrix were used to calculate the expected count rates in Bonner spheres spectrometer. These count rates were used as input and the respective spectrum was used as output during neural network training. After training the network was tested with the Bonner spheres count rates produced by a set of neutron spectra. This set contains data used during network training as well as data not used. Training and testing was carried out in the Mat lab program. To verify the network unfolding performance the original and unfolded spectra were compared using the χ 2 -test and the total fluence ratios. The use of Artificial Neural Networks to unfold neutron spectra in neutron spectrometry is an alternative procedure that overcomes the drawbacks associated in this ill-conditioned problem. (Author)

Accelerating deep neural network training with inconsistent stochastic gradient descent.

Science.gov (United States)

Wang, Linnan; Yang, Yi; Min, Renqiang; Chakradhar, Srimat

2017-09-01

Stochastic Gradient Descent (SGD) updates Convolutional Neural Network (CNN) with a noisy gradient computed from a random batch, and each batch evenly updates the network once in an epoch. This model applies the same training effort to each batch, but it overlooks the fact that the gradient variance, induced by Sampling Bias and Intrinsic Image Difference, renders different training dynamics on batches. In this paper, we develop a new training strategy for SGD, referred to as Inconsistent Stochastic Gradient Descent (ISGD) to address this problem. The core concept of ISGD is the inconsistent training, which dynamically adjusts the training effort w.r.t the loss. ISGD models the training as a stochastic process that gradually reduces down the mean of batch's loss, and it utilizes a dynamic upper control limit to identify a large loss batch on the fly. ISGD stays on the identified batch to accelerate the training with additional gradient updates, and it also has a constraint to penalize drastic parameter changes. ISGD is straightforward, computationally efficient and without requiring auxiliary memories. A series of empirical evaluations on real world datasets and networks demonstrate the promising performance of inconsistent training. Copyright © 2017 Elsevier Ltd. All rights reserved.

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....

Machine Learning Topological Invariants with Neural Networks

Science.gov (United States)

Zhang, Pengfei; Shen, Huitao; Zhai, Hui

2018-02-01

In this Letter we supervisedly train neural networks to distinguish different topological phases in the context of topological band insulators. After training with Hamiltonians of one-dimensional insulators with chiral symmetry, the neural network can predict their topological winding numbers with nearly 100% accuracy, even for Hamiltonians with larger winding numbers that are not included in the training data. These results show a remarkable success that the neural network can capture the global and nonlinear topological features of quantum phases from local inputs. By opening up the neural network, we confirm that the network does learn the discrete version of the winding number formula. We also make a couple of remarks regarding the role of the symmetry and the opposite effect of regularization techniques when applying machine learning to physical systems.

Training Excitatory-Inhibitory Recurrent Neural Networks for Cognitive Tasks: A Simple and Flexible Framework

Science.gov (United States)

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

Neural Network Ensembles

DEFF Research Database (Denmark)

Hansen, Lars Kai; Salamon, Peter

1990-01-01

We propose several means for improving the performance an training of neural networks for classification. We use crossvalidation as a tool for optimizing network parameters and architecture. We show further that the remaining generalization error can be reduced by invoking ensembles of similar...... networks....

Neural Networks in R Using the Stuttgart Neural Network Simulator: RSNNS

Directory of Open Access Journals (Sweden)

Christopher Bergmeir

2012-01-01

Full Text Available Neural networks are important standard machine learning procedures for classification and regression. We describe the R package RSNNS that provides a convenient interface to the popular Stuttgart Neural Network Simulator SNNS. The main features are (a encapsulation of the relevant SNNS parts in a C++ class, for sequential and parallel usage of different networks, (b accessibility of all of the SNNSalgorithmic functionality from R using a low-level interface, and (c a high-level interface for convenient, R-style usage of many standard neural network procedures. The package also includes functions for visualization and analysis of the models and the training procedures, as well as functions for data input/output from/to the original SNNSfile formats.

Neutron spectrometry using artificial neural networks

International Nuclear Information System (INIS)

Vega-Carrillo, Hector Rene; Martin Hernandez-Davila, Victor; Manzanares-Acuna, Eduardo; Mercado Sanchez, Gema A.; Pilar Iniguez de la Torre, Maria; Barquero, Raquel; Palacios, Francisco; Mendez Villafane, Roberto; Arteaga Arteaga, Tarcicio; Manuel Ortiz Rodriguez, Jose

2006-01-01

An artificial neural network has been designed to obtain neutron spectra from Bonner spheres spectrometer count rates. The neural network was trained using 129 neutron spectra. These include spectra from isotopic neutron sources; reference and operational spectra from accelerators and nuclear reactors, spectra based on mathematical functions as well as few energy groups and monoenergetic spectra. The spectra were transformed from lethargy to energy distribution and were re-binned to 31 energy groups using the MCNP 4C code. The re-binned spectra and the UTA4 response matrix were used to calculate the expected count rates in Bonner spheres spectrometer. These count rates were used as input and their respective spectra were used as output during the neural network training. After training, the network was tested with the Bonner spheres count rates produced by folding a set of neutron spectra with the response matrix. This set contains data used during network training as well as data not used. Training and testing was carried out using the Matlab ( R) program. To verify the network unfolding performance, the original and unfolded spectra were compared using the root mean square error. The use of artificial neural networks to unfold neutron spectra in neutron spectrometry is an alternative procedure that overcomes the drawbacks associated with this ill-conditioned problem

A Fast C++ Implementation of Neural Network Backpropagation Training Algorithm: Application to Bayesian Optimal Image Demosaicing

Directory of Open Access Journals (Sweden)

Yi-Qing Wang

2015-09-01

Full Text Available Recent years have seen a surge of interest in multilayer neural networks fueled by their successful applications in numerous image processing and computer vision tasks. In this article, we describe a C++ implementation of the stochastic gradient descent to train a multilayer neural network, where a fast and accurate acceleration of tanh(· is achieved with linear interpolation. As an example of application, we present a neural network able to deliver state-of-the-art performance in image demosaicing.

SAGRAD: A Program for Neural Network Training with Simulated Annealing and the Conjugate Gradient Method.

Science.gov (United States)

Bernal, Javier; Torres-Jimenez, Jose

2015-01-01

SAGRAD (Simulated Annealing GRADient), a Fortran 77 program for computing neural networks for classification using batch learning, is discussed. Neural network training in SAGRAD is based on a combination of simulated annealing and Møller's scaled conjugate gradient algorithm, the latter a variation of the traditional conjugate gradient method, better suited for the nonquadratic nature of neural networks. Different aspects of the implementation of the training process in SAGRAD are discussed, such as the efficient computation of gradients and multiplication of vectors by Hessian matrices that are required by Møller's algorithm; the (re)initialization of weights with simulated annealing required to (re)start Møller's algorithm the first time and each time thereafter that it shows insufficient progress in reaching a possibly local minimum; and the use of simulated annealing when Møller's algorithm, after possibly making considerable progress, becomes stuck at a local minimum or flat area of weight space. Outlines of the scaled conjugate gradient algorithm, the simulated annealing procedure and the training process used in SAGRAD are presented together with results from running SAGRAD on two examples of training data.

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

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.

Artificial Neural Network Modeling of an Inverse Fluidized Bed ...

African Journals Online (AJOL)

A Radial Basis Function neural network has been successfully employed for the modeling of the inverse fluidized bed reactor. In the proposed model, the trained neural network represents the kinetics of biological decomposition of pollutants in the reactor. The neural network has been trained with experimental data ...

Online Sequence Training of Recurrent Neural Networks with Connectionist Temporal Classification

OpenAIRE

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...

Analysis of neural networks through base functions

NARCIS (Netherlands)

van der Zwaag, B.J.; Slump, Cornelis H.; Spaanenburg, L.

Problem statement. Despite their success-story, neural networks have one major disadvantage compared to other techniques: the inability to explain comprehensively how a trained neural network reaches its output; neural networks are not only (incorrectly) seen as a "magic tool" but possibly even more

Reconstruction of sparse connectivity in neural networks from spike train covariances

International Nuclear Information System (INIS)

Pernice, Volker; Rotter, Stefan

2013-01-01

The inference of causation from correlation is in general highly problematic. Correspondingly, it is difficult to infer the existence of physical synaptic connections between neurons from correlations in their activity. Covariances in neural spike trains and their relation to network structure have been the subject of intense research, both experimentally and theoretically. The influence of recurrent connections on covariances can be characterized directly in linear models, where connectivity in the network is described by a matrix of linear coupling kernels. However, as indirect connections also give rise to covariances, the inverse problem of inferring network structure from covariances can generally not be solved unambiguously. Here we study to what degree this ambiguity can be resolved if the sparseness of neural networks is taken into account. To reconstruct a sparse network, we determine the minimal set of linear couplings consistent with the measured covariances by minimizing the L 1 norm of the coupling matrix under appropriate constraints. Contrary to intuition, after stochastic optimization of the coupling matrix, the resulting estimate of the underlying network is directed, despite the fact that a symmetric matrix of count covariances is used for inference. The performance of the new method is best if connections are neither exceedingly sparse, nor too dense, and it is easily applicable for networks of a few hundred nodes. Full coupling kernels can be obtained from the matrix of full covariance functions. We apply our method to networks of leaky integrate-and-fire neurons in an asynchronous–irregular state, where spike train covariances are well described by a linear model. (paper)

Application of a neural network for reflectance spectrum classification

Science.gov (United States)

Yang, Gefei; Gartley, Michael

2017-05-01

Traditional reflectance spectrum classification algorithms are based on comparing spectrum across the electromagnetic spectrum anywhere from the ultra-violet to the thermal infrared regions. These methods analyze reflectance on a pixel by pixel basis. Inspired by high performance that Convolution Neural Networks (CNN) have demonstrated in image classification, we applied a neural network to analyze directional reflectance pattern images. By using the bidirectional reflectance distribution function (BRDF) data, we can reformulate the 4-dimensional into 2 dimensions, namely incident direction × reflected direction × channels. Meanwhile, RIT's micro-DIRSIG model is utilized to simulate additional training samples for improving the robustness of the neural networks training. Unlike traditional classification by using hand-designed feature extraction with a trainable classifier, neural networks create several layers to learn a feature hierarchy from pixels to classifier and all layers are trained jointly. Hence, the our approach of utilizing the angular features are different to traditional methods utilizing spatial features. Although training processing typically has a large computational cost, simple classifiers work well when subsequently using neural network generated features. Currently, most popular neural networks such as VGG, GoogLeNet and AlexNet are trained based on RGB spatial image data. Our approach aims to build a directional reflectance spectrum based neural network to help us to understand from another perspective. At the end of this paper, we compare the difference among several classifiers and analyze the trade-off among neural networks parameters.

Application of artificial neural network in radiographic diagnosis

International Nuclear Information System (INIS)

Piraino, D.; Amartur, S.; Richmond, B.; Schils, J.; Belhobek, G.

1990-01-01

This paper reports on an artificial neural network trained to rate the likelihood of different bone neoplasms when given a standard description of a radiograph. A three-layer back propagation algorithm was trained with descriptions of examples of bone neoplasms obtained from standard radiographic textbooks. Fifteen bone neoplasms obtained from clinical material were used as unknowns to test the trained artificial neural network. The artificial neural network correctly rated the pathologic diagnosis as the most likely diagnosis in 10 of the 15 unknown cases

Deep Convolutional Neural Networks: Structure, Feature Extraction and Training

Directory of Open Access Journals (Sweden)

Namatēvs Ivars

2017-12-01

Full Text Available Deep convolutional neural networks (CNNs are aimed at processing data that have a known network like topology. They are widely used to recognise objects in images and diagnose patterns in time series data as well as in sensor data classification. The aim of the paper is to present theoretical and practical aspects of deep CNNs in terms of convolution operation, typical layers and basic methods to be used for training and learning. Some practical applications are included for signal and image classification. Finally, the present paper describes the proposed block structure of CNN for classifying crucial features from 3D sensor data.

Genetic Algorithm Optimized Neural Networks Ensemble as ...

African Journals Online (AJOL)

NJD

Improvements in neural network calibration models by a novel approach using neural network ensemble (NNE) for the simultaneous ... process by training a number of neural networks. .... Matlab® version 6.1 was employed for building principal component ... provide a fair simulation of calibration data set with some degree.

Training Feedforward Neural Networks Using Symbiotic Organisms Search Algorithm

Directory of Open Access Journals (Sweden)

Haizhou Wu

2016-01-01

Full Text Available Symbiotic organisms search (SOS is a new robust and powerful metaheuristic algorithm, which stimulates the symbiotic interaction strategies adopted by organisms to survive and propagate in the ecosystem. In the supervised learning area, it is a challenging task to present a satisfactory and efficient training algorithm for feedforward neural networks (FNNs. In this paper, SOS is employed as a new method for training FNNs. To investigate the performance of the aforementioned method, eight different datasets selected from the UCI machine learning repository are employed for experiment and the results are compared among seven metaheuristic algorithms. The results show that SOS performs better than other algorithms for training FNNs in terms of converging speed. It is also proven that an FNN trained by the method of SOS has better accuracy than most algorithms compared.

Advances in Artificial Neural Networks - Methodological Development and Application

Science.gov (United States)

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 ne...

Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?

OpenAIRE

Tajbakhsh, Nima; Shin, Jae Y.; Gurudu, Suryakanth R.; Hurst, R. Todd; Kendall, Christopher B.; Gotway, Michael B.; Liang, Jianming

2017-01-01

Training a deep convolutional neural network (CNN) from scratch is difficult because it requires a large amount of labeled training data and a great deal of expertise to ensure proper convergence. A promising alternative is to fine-tune a CNN that has been pre-trained using, for instance, a large set of labeled natural images. However, the substantial differences between natural and medical images may advise against such knowledge transfer. In this paper, we seek to answer the following centr...

Foreign currency rate forecasting using neural networks

Science.gov (United States)

Pandya, Abhijit S.; Kondo, Tadashi; Talati, Amit; Jayadevappa, Suryaprasad

2000-03-01

Neural networks are increasingly being used as a forecasting tool in many forecasting problems. This paper discusses the application of neural networks in predicting daily foreign exchange rates between the USD, GBP as well as DEM. We approach the problem from a time-series analysis framework - where future exchange rates are forecasted solely using past exchange rates. This relies on the belief that the past prices and future prices are very close related, and interdependent. We present the result of training a neural network with historical USD-GBP data. The methodology used in explained, as well as the training process. We discuss the selection of inputs to the network, and present a comparison of using the actual exchange rates and the exchange rate differences as inputs. Price and rate differences are the preferred way of training neural network in financial applications. Results of both approaches are present together for comparison. We show that the network is able to learn the trends in the exchange rate movements correctly, and present the results of the prediction over several periods of time.

Generating Seismograms with Deep Neural Networks

Science.gov (United States)

Krischer, L.; Fichtner, A.

2017-12-01

The recent surge of successful uses of deep neural networks in computer vision, speech recognition, and natural language processing, mainly enabled by the availability of fast GPUs and extremely large data sets, is starting to see many applications across all natural sciences. In seismology these are largely confined to classification and discrimination tasks. In this contribution we explore the use of deep neural networks for another class of problems: so called generative models.Generative modelling is a branch of statistics concerned with generating new observed data samples, usually by drawing from some underlying probability distribution. Samples with specific attributes can be generated by conditioning on input variables. In this work we condition on seismic source (mechanism and location) and receiver (location) parameters to generate multi-component seismograms.The deep neural networks are trained on synthetic data calculated with Instaseis (http://instaseis.net, van Driel et al. (2015)) and waveforms from the global ShakeMovie project (http://global.shakemovie.princeton.edu, Tromp et al. (2010)). The underlying radially symmetric or smoothly three dimensional Earth structures result in comparatively small waveform differences from similar events or at close receivers and the networks learn to interpolate between training data samples.Of particular importance is the chosen misfit functional. Generative adversarial networks (Goodfellow et al. (2014)) implement a system in which two networks compete: the generator network creates samples and the discriminator network distinguishes these from the true training examples. Both are trained in an adversarial fashion until the discriminator can no longer distinguish between generated and real samples. We show how this can be applied to seismograms and in particular how it compares to networks trained with more conventional misfit metrics. Last but not least we attempt to shed some light on the black-box nature of

Neural Networks

International Nuclear Information System (INIS)

Smith, Patrick I.

2003-01-01

Physicists use large detectors to measure particles created in high-energy collisions at particle accelerators. These detectors typically produce signals indicating either where ionization occurs along the path of the particle, or where energy is deposited by the particle. The data produced by these signals is fed into pattern recognition programs to try to identify what particles were produced, and to measure the energy and direction of these particles. Ideally, there are many techniques used in this pattern recognition software. One technique, neural networks, is particularly suitable for identifying what type of particle caused by a set of energy deposits. Neural networks can derive meaning from complicated or imprecise data, extract patterns, and detect trends that are too complex to be noticed by either humans or other computer related processes. To assist in the advancement of this technology, Physicists use a tool kit to experiment with several neural network techniques. The goal of this research is interface a neural network tool kit into Java Analysis Studio (JAS3), an application that allows data to be analyzed from any experiment. As the final result, a physicist will have the ability to train, test, and implement a neural network with the desired output while using JAS3 to analyze the results or output. Before an implementation of a neural network can take place, a firm understanding of what a neural network is and how it works is beneficial. A neural network is an artificial representation of the human brain that tries to simulate the learning process [5]. It is also important to think of the word artificial in that definition as computer programs that use calculations during the learning process. In short, a neural network learns by representative examples. Perhaps the easiest way to describe the way neural networks learn is to explain how the human brain functions. The human brain contains billions of neural cells that are responsible for processing

Neutron spectrum unfolding using neural networks

International Nuclear Information System (INIS)

Vega C, H.R.; Hernandez D, V.M.; Manzanares A, E.

2004-01-01

An artificial neural network has been designed to obtain the neutron spectra from the Bonner spheres spectrometer's count rates. The neural network was trained using a large set of neutron spectra compiled by the International Atomic Energy Agency. These include spectra from iso- topic neutron sources, reference and operational neutron spectra obtained from accelerators and nuclear reactors. The spectra were transformed from lethargy to energy distribution and were re-binned to 31 energy groups using the MCNP 4C code. Re-binned spectra and UTA4 matrix were used to calculate the expected count rates in Bonner spheres spectrometer. These count rates were used as input and correspondent spectrum was used as output during neural network training. The network has 7 input nodes, 56 neurons as hidden layer and 31 neurons in the output layer. After training the network was tested with the Bonner spheres count rates produced by twelve neutron spectra. The network allows unfolding the neutron spectrum from count rates measured with Bonner spheres. Good results are obtained when testing count rates belong to neutron spectra used during training, acceptable results are obtained for count rates obtained from actual neutron fields; however the network fails when count rates belong to monoenergetic neutron sources. (Author)

Fast neutron spectra determination by threshold activation detectors using neural networks

International Nuclear Information System (INIS)

Kardan, M.R.; Koohi-Fayegh, R.; Setayeshi, S.; Ghiassi-Nejad, M.

2004-01-01

Neural network method was used for fast neutron spectra unfolding in spectrometry by threshold activation detectors. The input layer of the neural networks consisted of 11 neurons for the specific activities of neutron-induced nuclear reaction products, while the output layers were fast neutron spectra which had been subdivided into 6, 8, 10, 12, 15 and 20 energy bins. Neural network training was performed by 437 fast neutron spectra and corresponding threshold activation detector readings. The trained neural network have been applied for unfolding 50 spectra, which were not in training sets and the results were compared with real spectra and unfolded spectra by SANDII. The best results belong to 10 energy bin spectra. The neural network was also trained by detector readings with 5% uncertainty and the response of the trained neural network to detector readings with 5%, 10%, 15%, 20%, 25% and 50% uncertainty was compared with real spectra. Neural network algorithm, in comparison with other unfolding methods, is very fast and needless to detector response matrix and any prior information about spectra and also the outputs have low sensitivity to uncertainty in the activity measurements. The results show that the neural network algorithm is useful when a fast response is required with reasonable accuracy

Neural network recognition of mammographic lesions

International Nuclear Information System (INIS)

Oldham, W.J.B.; Downes, P.T.; Hunter, V.

1987-01-01

A method for recognition of mammographic lesions through the use of neural networks is presented. Neural networks have exhibited the ability to learn the shape andinternal structure of patterns. Digitized mammograms containing circumscribed and stelate lesions were used to train a feedfoward synchronous neural network that self-organizes to stable attractor states. Encoding of data for submission to the network was accomplished by performing a fractal analysis of the digitized image. This results in scale invariant representation of the lesions. Results are discussed

Thermoelastic steam turbine rotor control based on neural network

Science.gov (United States)

Rzadkowski, Romuald; Dominiczak, Krzysztof; Radulski, Wojciech; Szczepanik, R.

2015-12-01

Considered here are Nonlinear Auto-Regressive neural networks with eXogenous inputs (NARX) as a mathematical model of a steam turbine rotor for controlling steam turbine stress on-line. In order to obtain neural networks that locate critical stress and temperature points in the steam turbine during transient states, an FE rotor model was built. This model was used to train the neural networks on the basis of steam turbine transient operating data. The training included nonlinearity related to steam turbine expansion, heat exchange and rotor material properties during transients. Simultaneous neural networks are algorithms which can be implemented on PLC controllers. This allows for the application neural networks to control steam turbine stress in industrial power plants.

Artificial Neural Network with Hardware Training and Hardware Refresh

Science.gov (United States)

Duong, Tuan A. (Inventor)

2003-01-01

A neural network circuit is provided having a plurality of circuits capable of charge storage. Also provided is a plurality of circuits each coupled to at least one of the plurality of charge storage circuits and constructed to generate an output in accordance with a neuron transfer function. Each of a plurality of circuits is coupled to one of the plurality of neuron transfer function circuits and constructed to generate a derivative of the output. A weight update circuit updates the charge storage circuits based upon output from the plurality of transfer function circuits and output from the plurality of derivative circuits. In preferred embodiments, separate training and validation networks share the same set of charge storage circuits and may operate concurrently. The validation network has a separate transfer function circuits each being coupled to the charge storage circuits so as to replicate the training network s coupling of the plurality of charge storage to the plurality of transfer function circuits. The plurality of transfer function circuits may be constructed each having a transconductance amplifier providing differential currents combined to provide an output in accordance with a transfer function. The derivative circuits may have a circuit constructed to generate a biased differential currents combined so as to provide the derivative of the transfer function.

LAI inversion from optical reflectance using a neural network trained with a multiple scattering model

Science.gov (United States)

Smith, James A.

1992-01-01

The inversion of the leaf area index (LAI) canopy parameter from optical spectral reflectance measurements is obtained using a backpropagation artificial neural network trained using input-output pairs generated by a multiple scattering reflectance model. The problem of LAI estimation over sparse canopies (LAI 1000 percent for low LAI. Minimization methods applied to merit functions constructed from differences between measured reflectances and predicted reflectances using multiple-scattering models are unacceptably sensitive to a good initial guess for the desired parameter. In contrast, the neural network reported generally yields absolute percentage errors of <30 percent when weighting coefficients trained on one soil type were applied to predicted canopy reflectance at a different soil background.

Estimation of Collapse Moment for Wall Thinned Elbows Using Fuzzy Neural Networks

International Nuclear Information System (INIS)

Na, Man Gyun; Kim, Jin Weon; Shin, Sun Ho; Kim, Koung Suk; Kang, Ki Soo

2004-01-01

In this work, the collapse moment due to wall-thinning defects is estimated by using fuzzy neural networks. The developed fuzzy neural networks have been applied to the numerical data obtained from the finite element analysis. Principal component analysis is used to preprocess the input signals into the fuzzy neural network to reduce the sensitivity to the input change and the fuzzy neural networks are trained by using the data set prepared for training (training data) and verified by using another data set different (independent) from the training data. Also, two fuzzy neural networks are trained for two data sets divided into the two classes of extrados and intrados defects, which is because they have different characteristics. The relative 2-sigma errors of the estimated collapse moment are 3.07% for the training data and 4.12% for the test data. It is known from this result that the fuzzy neural networks are sufficiently accurate to be used in the wall-thinning monitoring of elbows

Applications of neural network to numerical analyses

International Nuclear Information System (INIS)

Takeda, Tatsuoki; Fukuhara, Makoto; Ma, Xiao-Feng; Liaqat, Ali

1999-01-01

Applications of a multi-layer neural network to numerical analyses are described. We are mainly concerned with the computed tomography and the solution of differential equations. In both cases as the objective functions for the training process of the neural network we employed residuals of the integral equation or the differential equations. This is different from the conventional neural network training where sum of the squared errors of the output values is adopted as the objective function. For model problems both the methods gave satisfactory results and the methods are considered promising for some kind of problems. (author)

Neural-Network Object-Recognition Program

Science.gov (United States)

Spirkovska, L.; Reid, M. B.

1993-01-01

HONTIOR computer program implements third-order neural network exhibiting invariance under translation, change of scale, and in-plane rotation. Invariance incorporated directly into architecture of network. Only one view of each object needed to train network for two-dimensional-translation-invariant recognition of object. Also used for three-dimensional-transformation-invariant recognition by training network on only set of out-of-plane rotated views. Written in C language.

Artificial neural networks in neutron dosimetry

Energy Technology Data Exchange (ETDEWEB)

Vega C, H.R.; Hernandez D, V.M.; Manzanares A, E.; Mercado, G.A.; Perales M, W.A.; Robles R, J.A. [Unidades Academicas de Estudios Nucleares, UAZ, A.P. 336, 98000 Zacatecas (Mexico); Gallego, E.; Lorente, A. [Depto. de Ingenieria Nuclear, Universidad Politecnica de Madrid, (Spain)

2005-07-01

An artificial neural network has been designed to obtain the neutron doses using only the Bonner spheres spectrometer's count rates. Ambient, personal and effective neutron doses were included. 187 neutron spectra were utilized to calculate the Bonner count rates and the neutron doses. The spectra were transformed from lethargy to energy distribution and were re-binned to 31 energy groups using the MCNP 4C code. Re-binned spectra, UTA4 response matrix and fluence-to-dose coefficients were used to calculate the count rates in Bonner spheres spectrometer and the doses. Count rates were used as input and the respective doses were used as output during neural network training. Training and testing was carried out in Mat lab environment. The artificial neural network performance was evaluated using the {chi}{sup 2}- test, where the original and calculated doses were compared. The use of Artificial Neural Networks in neutron dosimetry is an alternative procedure that overcomes the drawbacks associated in this ill-conditioned problem. (Author)

Artificial neural networks in neutron dosimetry

International Nuclear Information System (INIS)

Vega C, H.R.; Hernandez D, V.M.; Manzanares A, E.; Mercado, G.A.; Perales M, W.A.; Robles R, J.A.; Gallego, E.; Lorente, A.

2005-01-01

An artificial neural network has been designed to obtain the neutron doses using only the Bonner spheres spectrometer's count rates. Ambient, personal and effective neutron doses were included. 187 neutron spectra were utilized to calculate the Bonner count rates and the neutron doses. The spectra were transformed from lethargy to energy distribution and were re-binned to 31 energy groups using the MCNP 4C code. Re-binned spectra, UTA4 response matrix and fluence-to-dose coefficients were used to calculate the count rates in Bonner spheres spectrometer and the doses. Count rates were used as input and the respective doses were used as output during neural network training. Training and testing was carried out in Mat lab environment. The artificial neural network performance was evaluated using the χ 2 - test, where the original and calculated doses were compared. The use of Artificial Neural Networks in neutron dosimetry is an alternative procedure that overcomes the drawbacks associated in this ill-conditioned problem. (Author)

Training algorithms evaluation for artificial neural network to temporal prediction of photovoltaic generation

International Nuclear Information System (INIS)

Arantes Monteiro, Raul Vitor; Caixeta Guimarães, Geraldo; Rocio Castillo, Madeleine; Matheus Moura, Fabrício Augusto; Tamashiro, Márcio Augusto

2016-01-01

Current energy policies are encouraging the connection of power generation based on low-polluting technologies, mainly those using renewable sources, to distribution networks. Hence, it becomes increasingly important to understand technical challenges, facing high penetration of PV systems at the grid, especially considering the effects of intermittence of this source on the power quality, reliability and stability of the electric distribution system. This fact can affect the distribution networks on which they are attached causing overvoltage, undervoltage and frequency oscillations. In order to predict these disturbs, artificial neural networks are used. This article aims to analyze 3 training algorithms used in artificial neural networks for temporal prediction of the generated active power thru photovoltaic panels. As a result it was concluded that the algorithm with the best performance among the 3 analyzed was the Levenberg-Marquadrt.

On the use of harmony search algorithm in the training of wavelet neural networks

Science.gov (United States)

Lai, Kee Huong; Zainuddin, Zarita; Ong, Pauline

2015-10-01

Wavelet neural networks (WNNs) are a class of feedforward neural networks that have been used in a wide range of industrial and engineering applications to model the complex relationships between the given inputs and outputs. The training of WNNs involves the configuration of the weight values between neurons. The backpropagation training algorithm, which is a gradient-descent method, can be used for this training purpose. Nonetheless, the solutions found by this algorithm often get trapped at local minima. In this paper, a harmony search-based algorithm is proposed for the training of WNNs. The training of WNNs, thus can be formulated as a continuous optimization problem, where the objective is to maximize the overall classification accuracy. Each candidate solution proposed by the harmony search algorithm represents a specific WNN architecture. In order to speed up the training process, the solution space is divided into disjoint partitions during the random initialization step of harmony search algorithm. The proposed training algorithm is tested onthree benchmark problems from the UCI machine learning repository, as well as one real life application, namely, the classification of electroencephalography signals in the task of epileptic seizure detection. The results obtained show that the proposed algorithm outperforms the traditional harmony search algorithm in terms of overall classification accuracy.

Neural network segmentation of magnetic resonance images

International Nuclear Information System (INIS)

Frederick, B.

1990-01-01

Neural networks are well adapted to the task of grouping input patterns into subsets which share some similarity. Moreover, once trained, they can generalize their classification rules to classify new data sets. Sets of pixel intensities from magnetic resonance (MR) images provide a natural input to a neural network; by varying imaging parameters, MR images can reflect various independent physical parameters of tissues in their pixel intensities. A neural net can then be trained to classify physically similar tissue types based on sets of pixel intensities resulting from different imaging studies on the same subject. This paper reports that a neural network classifier for image segmentation was implanted on a Sun 4/60, and was tested on the task of classifying tissues of canine head MR images. Four images of a transaxial slice with different imaging sequences were taken as input to the network (three spin-echo images and an inversion recovery image). The training set consisted of 691 representative samples of gray matter, white matter, cerebrospinal fluid, bone, and muscle preclassified by a neuroscientist. The network was trained using a fast backpropagation algorithm to derive the decision criteria to classify any location in the image by its pixel intensities, and the image was subsequently segmented by the classifier

Application of neural networks in coastal engineering

Digital Repository Service at National Institute of Oceanography (India)

Mandal, S.

the neural network attractive. A neural network is an information processing system modeled on the structure of the dynamic process. It can solve the complex/nonlinear problems quickly once trained by operating on problems using an interconnected number...

Evolutionary Algorithms For Neural Networks Binary And Real Data Classification

Directory of Open Access Journals (Sweden)

Dr. Hanan A.R. Akkar

2015-08-01

Full Text Available Artificial neural networks are complex networks emulating the way human rational neurons process data. They have been widely used generally in prediction clustering classification and association. The training algorithms that used to determine the network weights are almost the most important factor that influence the neural networks performance. Recently many meta-heuristic and Evolutionary algorithms are employed to optimize neural networks weights to achieve better neural performance. This paper aims to use recently proposed algorithms for optimizing neural networks weights comparing these algorithms performance with other classical meta-heuristic algorithms used for the same purpose. However to evaluate the performance of such algorithms for training neural networks we examine such algorithms to classify four opposite binary XOR clusters and classification of continuous real data sets such as Iris and Ecoli.

Modeling of steam generator in nuclear power plant using neural network ensemble

International Nuclear Information System (INIS)

Lee, S. K.; Lee, E. C.; Jang, J. W.

2003-01-01

Neural network is now being used in modeling the steam generator is known to be difficult due to the reverse dynamics. However, Neural network is prone to the problem of overfitting. This paper investigates the use of neural network combining methods to model steam generator water level and compares with single neural network. The results show that neural network ensemble is effective tool which can offer improved generalization, lower dependence of the training set and reduced training time

Neural networks and orbit control in accelerators

International Nuclear Information System (INIS)

Bozoki, E.; Friedman, A.

1994-01-01

An overview of the architecture, workings and training of Neural Networks is given. We stress the aspects which are important for the use of Neural Networks for orbit control in accelerators and storage rings, especially its ability to cope with the nonlinear behavior of the orbit response to 'kicks' and the slow drift in the orbit response during long-term operation. Results obtained for the two NSLS storage rings with several network architectures and various training methods for each architecture are given

Illusory Motion Reproduced by Deep Neural Networks Trained for Prediction.

Science.gov (United States)

Watanabe, Eiji; Kitaoka, Akiyoshi; Sakamoto, Kiwako; Yasugi, Masaki; Tanaka, Kenta

2018-01-01

The cerebral cortex predicts visual motion to adapt human behavior to surrounding objects moving in real time. Although the underlying mechanisms are still unknown, predictive coding is one of the leading theories. Predictive coding assumes that the brain's internal models (which are acquired through learning) predict the visual world at all times and that errors between the prediction and the actual sensory input further refine the internal models. In the past year, deep neural networks based on predictive coding were reported for a video prediction machine called PredNet. If the theory substantially reproduces the visual information processing of the cerebral cortex, then PredNet can be expected to represent the human visual perception of motion. In this study, PredNet was trained with natural scene videos of the self-motion of the viewer, and the motion prediction ability of the obtained computer model was verified using unlearned videos. We found that the computer model accurately predicted the magnitude and direction of motion of a rotating propeller in unlearned videos. Surprisingly, it also represented the rotational motion for illusion images that were not moving physically, much like human visual perception. While the trained network accurately reproduced the direction of illusory rotation, it did not detect motion components in negative control pictures wherein people do not perceive illusory motion. This research supports the exciting idea that the mechanism assumed by the predictive coding theory is one of basis of motion illusion generation. Using sensory illusions as indicators of human perception, deep neural networks are expected to contribute significantly to the development of brain research.

Multi-modular neural networks for the classification of e+e- hadronic events

International Nuclear Information System (INIS)

Proriol, J.

1994-01-01

Some multi-modular neural network methods of classifying e + e - hadronic events are presented. We compare the performances of the following neural networks: MLP (multilayer perceptron), MLP and LVQ (learning vector quantization) trained sequentially, and MLP and RBF (radial basis function) trained sequentially. We introduce a MLP-RBF cooperative neural network. Our last study is a multi-MLP neural network. (orig.)

Using neural networks to describe tracer correlations

Directory of Open Access Journals (Sweden)

D. J. Lary

2004-01-01

Full Text Available Neural networks are ideally suited to describe the spatial and temporal dependence of tracer-tracer correlations. The neural network performs well even in regions where the correlations are less compact and normally a family of correlation curves would be required. For example, the CH4-N2O correlation can be well described using a neural network trained with the latitude, pressure, time of year, and methane volume mixing ratio (v.m.r.. In this study a neural network using Quickprop learning and one hidden layer with eight nodes was able to reproduce the CH4-N2O correlation with a correlation coefficient between simulated and training values of 0.9995. Such an accurate representation of tracer-tracer correlations allows more use to be made of long-term datasets to constrain chemical models. Such as the dataset from the Halogen Occultation Experiment (HALOE which has continuously observed CH4  (but not N2O from 1991 till the present. The neural network Fortran code used is available for download.

Deep learning quick reference useful hacks for training and optimizing deep neural networks with TensorFlow and Keras

CERN Document Server

Bernico, Michael

2018-01-01

This book is a practical guide to applying deep neural networks including MLPs, CNNs, LSTMs, and more in Keras and TensorFlow. Packed with useful hacks to solve real-world challenges along with the supported math and theory around each topic, this book will be a quick reference for training and optimize your deep neural networks.

Analysis of neural networks in terms of domain functions

NARCIS (Netherlands)

van der Zwaag, B.J.; Slump, Cornelis H.; Spaanenburg, Lambert

Despite their success-story, artificial neural networks have one major disadvantage compared to other techniques: the inability to explain comprehensively how a trained neural network reaches its output; neural networks are not only (incorrectly) seen as a "magic tool" but possibly even more as a

One weird trick for parallelizing convolutional neural networks

OpenAIRE

Krizhevsky, Alex

2014-01-01

I present a new way to parallelize the training of convolutional neural networks across multiple GPUs. The method scales significantly better than all alternatives when applied to modern convolutional neural networks.

Wind power forecast using wavelet neural network trained by improved Clonal selection algorithm

International Nuclear Information System (INIS)

Chitsaz, Hamed; Amjady, Nima; Zareipour, Hamidreza

2015-01-01

Highlights: • Presenting a Morlet wavelet neural network for wind power forecasting. • Proposing improved Clonal selection algorithm for training the model. • Applying Maximum Correntropy Criterion to evaluate the training performance. • Extensive testing of the proposed wind power forecast method on real-world data. - Abstract: With the integration of wind farms into electric power grids, an accurate wind power prediction is becoming increasingly important for the operation of these power plants. In this paper, a new forecasting engine for wind power prediction is proposed. The proposed engine has the structure of Wavelet Neural Network (WNN) with the activation functions of the hidden neurons constructed based on multi-dimensional Morlet wavelets. This forecast engine is trained by a new improved Clonal selection algorithm, which optimizes the free parameters of the WNN for wind power prediction. Furthermore, Maximum Correntropy Criterion (MCC) has been utilized instead of Mean Squared Error as the error measure in training phase of the forecasting model. The proposed wind power forecaster is tested with real-world hourly data of system level wind power generation in Alberta, Canada. In order to demonstrate the efficiency of the proposed method, it is compared with several other wind power forecast techniques. The obtained results confirm the validity of the developed approach

An Evolutionary Optimization Framework for Neural Networks and Neuromorphic Architectures

Energy Technology Data Exchange (ETDEWEB)

Schuman, Catherine D [ORNL; Plank, James [University of Tennessee (UT); Disney, Adam [University of Tennessee (UT); Reynolds, John [University of Tennessee (UT)

2016-01-01

As new neural network and neuromorphic architectures are being developed, new training methods that operate within the constraints of the new architectures are required. Evolutionary optimization (EO) is a convenient training method for new architectures. In this work, we review a spiking neural network architecture and a neuromorphic architecture, and we describe an EO training framework for these architectures. We present the results of this training framework on four classification data sets and compare those results to other neural network and neuromorphic implementations. We also discuss how this EO framework may be extended to other architectures.

Efficient Neural Network Modeling for Flight and Space Dynamics Simulation

Directory of Open Access Journals (Sweden)

Ayman Hamdy Kassem

2011-01-01

Full Text Available This paper represents an efficient technique for neural network modeling of flight and space dynamics simulation. The technique will free the neural network designer from guessing the size and structure for the required neural network model and will help to minimize the number of neurons. For linear flight/space dynamics systems, the technique can find the network weights and biases directly by solving a system of linear equations without the need for training. Nonlinear flight dynamic systems can be easily modeled by training its linearized models keeping the same network structure. The training is fast, as it uses the linear system knowledge to speed up the training process. The technique is tested on different flight/space dynamic models and showed promising results.

Accident scenario diagnostics with neural networks

International Nuclear Information System (INIS)

Guo, Z.

1992-01-01

Nuclear power plants are very complex systems. The diagnoses of transients or accident conditions is very difficult because a large amount of information, which is often noisy, or intermittent, or even incomplete, need to be processed in real time. To demonstrate their potential application to nuclear power plants, neural networks axe used to monitor the accident scenarios simulated by the training simulator of TVA's Watts Bar Nuclear Power Plant. A self-organization network is used to compress original data to reduce the total number of training patterns. Different accident scenarios are closely related to different key parameters which distinguish one accident scenario from another. Therefore, the accident scenarios can be monitored by a set of small size neural networks, called modular networks, each one of which monitors only one assigned accident scenario, to obtain fast training and recall. Sensitivity analysis is applied to select proper input variables for modular networks

Neutron spectra unfolding in Bonner spheres spectrometry using neural networks

International Nuclear Information System (INIS)

Kardan, M.R.; Setayeshi, S.; Koohi-Fayegh, R.; Ghiassi-Nejad, M.

2003-01-01

The neural network method has been used for the unfolding of neutron spectra in neutron spectrometry by Bonner spheres. A back propagation algorithm was used for training of neural networks 4mm x 4 mm bare LiI(Eu) and in a polyethylene sphere set: 2, 3, 4, 5, 6, 7, 8, 10, 12, 18 inch diameter have been used for unfolding of neutron spectra. Neural networks were trained by 199 sets of neutron spectra, which were subdivided into 6, 8, 10, 12, 15 and 20 energy bins and for each of them an appropriate neural network was designed and trained. The validation was performed by the 21 sets of neutron spectra. A neural network with 10 energy bins which had a mean value of error of 6% for dose equivalent estimation of spectra in the validation set showed the best results. The obtained results show that neural networks can be applied as an effective method for unfolding neutron spectra especially when the main target is neutron dosimetry. (author)

Optical-Correlator Neural Network Based On Neocognitron

Science.gov (United States)

Chao, Tien-Hsin; Stoner, William W.

1994-01-01

Multichannel optical correlator implements shift-invariant, high-discrimination pattern-recognizing neural network based on paradigm of neocognitron. Selected as basic building block of this neural network because invariance under shifts is inherent advantage of Fourier optics included in optical correlators in general. Neocognitron is conceptual electronic neural-network model for recognition of visual patterns. Multilayer processing achieved by iteratively feeding back output of feature correlator to input spatial light modulator and updating Fourier filters. Neural network trained by use of characteristic features extracted from target images. Multichannel implementation enables parallel processing of large number of selected features.

Tensor Basis Neural Network v. 1.0 (beta)

Energy Technology Data Exchange (ETDEWEB)

2017-03-28

This software package can be used to build, train, and test a neural network machine learning model. The neural network architecture is specifically designed to embed tensor invariance properties by enforcing that the model predictions sit on an invariant tensor basis. This neural network architecture can be used in developing constitutive models for applications such as turbulence modeling, materials science, and electromagnetism.

Stability prediction of berm breakwater using neural network

Digital Repository Service at National Institute of Oceanography (India)

Mandal, S.; Rao, S.; Manjunath, Y.R.

In the present study, an artificial neural network method has been applied to predict the stability of berm breakwaters. Four neural network models are constructed based on the parameters which influence the stability of breakwater. Training...

Illusory Motion Reproduced by Deep Neural Networks Trained for Prediction

Directory of Open Access Journals (Sweden)

Eiji Watanabe

2018-03-01

Full Text Available The cerebral cortex predicts visual motion to adapt human behavior to surrounding objects moving in real time. Although the underlying mechanisms are still unknown, predictive coding is one of the leading theories. Predictive coding assumes that the brain's internal models (which are acquired through learning predict the visual world at all times and that errors between the prediction and the actual sensory input further refine the internal models. In the past year, deep neural networks based on predictive coding were reported for a video prediction machine called PredNet. If the theory substantially reproduces the visual information processing of the cerebral cortex, then PredNet can be expected to represent the human visual perception of motion. In this study, PredNet was trained with natural scene videos of the self-motion of the viewer, and the motion prediction ability of the obtained computer model was verified using unlearned videos. We found that the computer model accurately predicted the magnitude and direction of motion of a rotating propeller in unlearned videos. Surprisingly, it also represented the rotational motion for illusion images that were not moving physically, much like human visual perception. While the trained network accurately reproduced the direction of illusory rotation, it did not detect motion components in negative control pictures wherein people do not perceive illusory motion. This research supports the exciting idea that the mechanism assumed by the predictive coding theory is one of basis of motion illusion generation. Using sensory illusions as indicators of human perception, deep neural networks are expected to contribute significantly to the development of brain research.

Neural network based multiscale image restoration approach

Science.gov (United States)

de Castro, Ana Paula A.; da Silva, José D. S.

2007-02-01

This paper describes a neural network based multiscale image restoration approach. Multilayer perceptrons are trained with artificial images of degraded gray level circles, in an attempt to make the neural network learn inherent space relations of the degraded pixels. The present approach simulates the degradation by a low pass Gaussian filter blurring operation and the addition of noise to the pixels at pre-established rates. The training process considers the degraded image as input and the non-degraded image as output for the supervised learning process. The neural network thus performs an inverse operation by recovering a quasi non-degraded image in terms of least squared. The main difference of the approach to existing ones relies on the fact that the space relations are taken from different scales, thus providing relational space data to the neural network. The approach is an attempt to come up with a simple method that leads to an optimum solution to the problem. Considering different window sizes around a pixel simulates the multiscale operation. In the generalization phase the neural network is exposed to indoor, outdoor, and satellite degraded images following the same steps use for the artificial circle image.

Drift chamber tracking with neural networks

International Nuclear Information System (INIS)

Lindsey, C.S.; Denby, B.; Haggerty, H.

1992-10-01

We discuss drift chamber tracking with a commercial log VLSI neural network chip. Voltages proportional to the drift times in a 4-layer drift chamber were presented to the Intel ETANN chip. The network was trained to provide the intercept and slope of straight tracks traversing the chamber. The outputs were recorded and later compared off line to conventional track fits. Two types of network architectures were studied. Applications of neural network tracking to high energy physics detector triggers is discussed

Mass reconstruction with a neural network

International Nuclear Information System (INIS)

Loennblad, L.; Peterson, C.; Roegnvaldsson, T.

1992-01-01

A feed-forward neural network method is developed for reconstructing the invariant mass of hadronic jets appearing in a calorimeter. The approach is illustrated in W→qanti q, where W-bosons are produced in panti p reactions at SPS collider energies. The neural network method yields results that are superior to conventional methods. This neural network application differs from the classification ones in the sense that an analog number (the mass) is computed by the network, rather than a binary decision being made. As a by-product our application clearly demonstrates the need for using 'intelligent' variables in instances when the amount of training instances is limited. (orig.)

Prediction of proteasome cleavage motifs by neural networks

DEFF Research Database (Denmark)

Kesimir, C.; Nussbaum, A.K.; Schild, H.

2002-01-01

physiological conditions. Our algorithm has been trained not only on in vitro data, but also on MHC Class I ligand data, which reflect a combination of immunoproteasome and constitutive proteasome specificity. This feature, together with the use of neural networks, a non-linear classification technique, make...... the prediction of MHC Class I ligand boundaries more accurate: 65% of the cleavage sites and 85% of the non-cleavage sites are correctly determined. Moreover, we show that the neural networks trained on the constitutive proteasome data learns a specificity that differs from that of the networks trained on MHC...

Analysis Resilient Algorithm on Artificial Neural Network Backpropagation

Science.gov (United States)

Saputra, Widodo; Tulus; Zarlis, Muhammad; Widia Sembiring, Rahmat; Hartama, Dedy

2017-12-01

Prediction required by decision makers to anticipate future planning. Artificial Neural Network (ANN) Backpropagation is one of method. This method however still has weakness, for long training time. This is a reason to improve a method to accelerate the training. One of Artificial Neural Network (ANN) Backpropagation method is a resilient method. Resilient method of changing weights and bias network with direct adaptation process of weighting based on local gradient information from every learning iteration. Predicting data result of Istanbul Stock Exchange training getting better. Mean Square Error (MSE) value is getting smaller and increasing accuracy.

EEG signal classification using PSO trained RBF neural network for epilepsy identification

Directory of Open Access Journals (Sweden)

Sandeep Kumar Satapathy

Full Text Available The electroencephalogram (EEG is a low amplitude signal generated in the brain, as a result of information flow during the communication of several neurons. Hence, careful analysis of these signals could be useful in understanding many human brain disorder diseases. One such disease topic is epileptic seizure identification, which can be identified via a classification process of the EEG signal after preprocessing with the discrete wavelet transform (DWT. To classify the EEG signal, we used a radial basis function neural network (RBFNN. As shown herein, the network can be trained to optimize the mean square error (MSE by using a modified particle swarm optimization (PSO algorithm. The key idea behind the modification of PSO is to introduce a method to overcome the problem of slow searching in and around the global optimum solution. The effectiveness of this procedure was verified by an experimental analysis on a benchmark dataset which is publicly available. The result of our experimental analysis revealed that the improvement in the algorithm is significant with respect to RBF trained by gradient descent and canonical PSO. Here, two classes of EEG signals were considered: the first being an epileptic and the other being non-epileptic. The proposed method produced a maximum accuracy of 99% as compared to the other techniques. Keywords: Electroencephalography, Radial basis function neural network, Particle swarm optimization, Discrete wavelet transform, Machine learning

Landslide Susceptibility Index Determination Using Aritificial Neural Network

Science.gov (United States)

Kawabata, D.; Bandibas, J.; Urai, M.

2004-12-01

The occurrence of landslide is the result of the interaction of complex and diverse environmental factors. The geomorphic features, rock types and geologic structure are especially important base factors of the landslide occurrence. Generating landslide susceptibility index by defining the relationship between landslide occurrence and that base factors using conventional mathematical and statistical methods is very difficult and inaccurate. This study focuses on generating landslide susceptibility index using artificial neural networks in Southern Japanese Alps. The training data are geomorphic (e.g. altitude, slope and aspect) and geologic parameters (e.g. rock type, distance from geologic boundary and geologic dip-strike angle) and landslides. Artificial neural network structure and training scheme are formulated to generate the index. Data from areas with and without landslide occurrences are used to train the network. The network is trained to output 1 when the input data are from areas with landslides and 0 when no landslide occurred. The trained network generates an output ranging from 0 to 1 reflecting the possibility of landslide occurrence based on the inputted data. Output values nearer to 1 means higher possibility of landslide occurrence. The artificial neural network model is incorporated into the GIS software to generate a landslide susceptibility map.

Using function approximation to determine neural network accuracy

International Nuclear Information System (INIS)

Wichman, R.F.; Alexander, J.

2013-01-01

Many, if not most, control processes demonstrate nonlinear behavior in some portion of their operating range and the ability of neural networks to model non-linear dynamics makes them very appealing for control. Control of high reliability safety systems, and autonomous control in process or robotic applications, however, require accurate and consistent control and neural networks are only approximators of various functions so their degree of approximation becomes important. In this paper, the factors affecting the ability of a feed-forward back-propagation neural network to accurately approximate a non-linear function are explored. Compared to pattern recognition using a neural network for function approximation provides an easy and accurate method for determining the network's accuracy. In contrast to other techniques, we show that errors arising in function approximation or curve fitting are caused by the neural network itself rather than scatter in the data. A method is proposed that provides improvements in the accuracy achieved during training and resulting ability of the network to generalize after training. Binary input vectors provided a more accurate model than with scalar inputs and retraining using a small number of the outlier x,y pairs improved generalization. (author)

Towards dropout training for convolutional neural networks.

Science.gov (United States)

Wu, Haibing; Gu, Xiaodong

2015-11-01

Recently, dropout has seen increasing use in deep learning. For deep convolutional neural networks, dropout is known to work well in fully-connected layers. However, its effect in convolutional and pooling layers is still not clear. This paper demonstrates that max-pooling dropout is equivalent to randomly picking activation based on a multinomial distribution at training time. In light of this insight, we advocate employing our proposed probabilistic weighted pooling, instead of commonly used max-pooling, to act as model averaging at test time. Empirical evidence validates the superiority of probabilistic weighted pooling. We also empirically show that the effect of convolutional dropout is not trivial, despite the dramatically reduced possibility of over-fitting due to the convolutional architecture. Elaborately designing dropout training simultaneously in max-pooling and fully-connected layers, we achieve state-of-the-art performance on MNIST, and very competitive results on CIFAR-10 and CIFAR-100, relative to other approaches without data augmentation. Finally, we compare max-pooling dropout and stochastic pooling, both of which introduce stochasticity based on multinomial distributions at pooling stage. Copyright © 2015 Elsevier Ltd. All rights reserved.

Potential usefulness of an artificial neural network for assessing ventricular size

International Nuclear Information System (INIS)

Fukuda, Haruyuki; Nakajima, Hideyuki; Usuki, Noriaki; Saiwai, Shigeo; Miyamoto, Takeshi; Inoue, Yuichi; Onoyama, Yasuto.

1995-01-01

An artificial neural network approach was applied to assess ventricular size from computed tomograms. Three layer, feed-forward neural networks with a back propagation algorithm were designed to distinguish between three degree of enlargement of the ventricles on the basis of patient's age and six items of computed tomographic information. Data for training and testing the neural network were created with computed tomograms of the brains selected at random from daily examinations. Four radiologists decided by mutual consent subjectively based on their experience whether the ventricles were within normal limits, slightly enlarged, or enlarged for the patient's age. The data for training was obtained from 38 patients. The data for testing was obtained from 47 other patients. The performance of the neural network trained using the data for training was evaluated by the rate of correct answers to the data for testing. The valid solution ratio to response of the test data obtained from the trained neural networks was more than 90% for all conditions in this study. The solutions were completely valid in the neural networks with two or three units at the hidden layer with 2,200 learning iterations, and with two units at the hidden layer with 11,000 learning iterations. The squared error decreased remarkably in the range from 0 to 500 learning iterations, and was close to a contrast over two thousand learning iterations. The neural network with a hidden layer having two or three units showed high decision performance. The preliminary results strongly suggest that the neural network approach has potential utility in computer-aided estimation of enlargement of the ventricles. (author)

Noise Analysis studies with neural networks

International Nuclear Information System (INIS)

Seker, S.; Ciftcioglu, O.

1996-01-01

Noise analysis studies with neural network are aimed. Stochastic signals at the input of the network are used to obtain an algorithmic multivariate stochastic signal modeling. To this end, lattice modeling of a stochastic signal is performed to obtain backward residual noise sources which are uncorrelated among themselves. There are applied together with an additional input to the network to obtain an algorithmic model which is used for signal detection for early failure in plant monitoring. The additional input provides the information to the network to minimize the difference between the signal and the network's one-step-ahead prediction. A stochastic algorithm is used for training where the errors reflecting the measurement error during the training are also modelled so that fast and consistent convergence of network's weights is obtained. The lattice structure coupled to neural network investigated with measured signals from an actual power plant. (authors)

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...

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

Science.gov (United States)

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.

Can surgical simulation be used to train detection and classification of neural networks?

Science.gov (United States)

Zisimopoulos, Odysseas; Flouty, Evangello; Stacey, Mark; Muscroft, Sam; Giataganas, Petros; Nehme, Jean; Chow, Andre; Stoyanov, Danail

2017-10-01

Computer-assisted interventions (CAI) aim to increase the effectiveness, precision and repeatability of procedures to improve surgical outcomes. The presence and motion of surgical tools is a key information input for CAI surgical phase recognition algorithms. Vision-based tool detection and recognition approaches are an attractive solution and can be designed to take advantage of the powerful deep learning paradigm that is rapidly advancing image recognition and classification. The challenge for such algorithms is the availability and quality of labelled data used for training. In this Letter, surgical simulation is used to train tool detection and segmentation based on deep convolutional neural networks and generative adversarial networks. The authors experiment with two network architectures for image segmentation in tool classes commonly encountered during cataract surgery. A commercially-available simulator is used to create a simulated cataract dataset for training models prior to performing transfer learning on real surgical data. To the best of authors' knowledge, this is the first attempt to train deep learning models for surgical instrument detection on simulated data while demonstrating promising results to generalise on real data. Results indicate that simulated data does have some potential for training advanced classification methods for CAI systems.

Nonlinear programming with feedforward neural networks.

Energy Technology Data Exchange (ETDEWEB)

Reifman, J.

1999-06-02

We provide a practical and effective method for solving constrained optimization problems by successively training a multilayer feedforward neural network in a coupled neural-network/objective-function representation. Nonlinear programming problems are easily mapped into this representation which has a simpler and more transparent method of solution than optimization performed with Hopfield-like networks and poses very mild requirements on the functions appearing in the problem. Simulation results are illustrated and compared with an off-the-shelf optimization tool.

Direct adaptive control using feedforward neural networks

OpenAIRE

Cajueiro, Daniel Oliveira; Hemerly, Elder Moreira

2003-01-01

ABSTRACT: This paper proposes a new scheme for direct neural adaptive control that works efficiently employing only one neural network, used for simultaneously identifying and controlling the plant. The idea behind this structure of adaptive control is to compensate the control input obtained by a conventional feedback controller. The neural network training process is carried out by using two different techniques: backpropagation and extended Kalman filter algorithm. Additionally, the conver...

PREDIKSI FOREX MENGGUNAKAN MODEL NEURAL NETWORK

Directory of Open Access Journals (Sweden)

R. Hadapiningradja Kusumodestoni

2015-11-01

Full Text Available ABSTRAK Prediksi adalah salah satu teknik yang paling penting dalam menjalankan bisnis forex. Keputusan dalam memprediksi adalah sangatlah penting, karena dengan prediksi dapat membantu mengetahui nilai forex di waktu tertentu kedepan sehingga dapat mengurangi resiko kerugian. Tujuan dari penelitian ini dimaksudkan memprediksi bisnis fores menggunakan model neural network dengan data time series per 1 menit untuk mengetahui nilai akurasi prediksi sehingga dapat mengurangi resiko dalam menjalankan bisnis forex. Metode penelitian pada penelitian ini meliputi metode pengumpulan data kemudian dilanjutkan ke metode training, learning, testing menggunakan neural network. Setelah di evaluasi hasil penelitian ini menunjukan bahwa penerapan algoritma Neural Network mampu untuk memprediksi forex dengan tingkat akurasi prediksi 0.431 +/- 0.096 sehingga dengan prediksi ini dapat membantu mengurangi resiko dalam menjalankan bisnis forex. Kata kunci: prediksi, forex, neural network.

Application of neural networks to seismic active control

International Nuclear Information System (INIS)

Tang, Yu.

1995-01-01

An exploratory study on seismic active control using an artificial neural network (ANN) is presented in which a singledegree-of-freedom (SDF) structural system is controlled by a trained neural network. A feed-forward neural network and the backpropagation training method are used in the study. In backpropagation training, the learning rate is determined by ensuring the decrease of the error function at each training cycle. The training patterns for the neural net are generated randomly. Then, the trained ANN is used to compute the control force according to the control algorithm. The control strategy proposed herein is to apply the control force at every time step to destroy the build-up of the system response. The ground motions considered in the simulations are the N21E and N69W components of the Lake Hughes No. 12 record that occurred in the San Fernando Valley in California on February 9, 1971. Significant reduction of the structural response by one order of magnitude is observed. Also, it is shown that the proposed control strategy has the ability to reduce the peak that occurs during the first few cycles of the time history. These promising results assert the potential of applying ANNs to active structural control under seismic loads

Interacting neural networks

Science.gov (United States)

Metzler, R.; Kinzel, W.; Kanter, I.

2000-08-01

Several scenarios of interacting neural networks which are trained either in an identical or in a competitive way are solved analytically. In the case of identical training each perceptron receives the output of its neighbor. The symmetry of the stationary state as well as the sensitivity to the used training algorithm are investigated. Two competitive perceptrons trained on mutually exclusive learning aims and a perceptron which is trained on the opposite of its own output are examined analytically. An ensemble of competitive perceptrons is used as decision-making algorithms in a model of a closed market (El Farol Bar problem or the Minority Game. In this game, a set of agents who have to make a binary decision is considered.); each network is trained on the history of minority decisions. This ensemble of perceptrons relaxes to a stationary state whose performance can be better than random.

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.

Neural-network-directed alignment of optical systems using the laser-beam spatial filter as an example

Science.gov (United States)

Decker, Arthur J.; Krasowski, Michael J.; Weiland, Kenneth E.

1993-01-01

This report describes an effort at NASA Lewis Research Center to use artificial neural networks to automate the alignment and control of optical measurement systems. Specifically, it addresses the use of commercially available neural network software and hardware to direct alignments of the common laser-beam-smoothing spatial filter. The report presents a general approach for designing alignment records and combining these into training sets to teach optical alignment functions to neural networks and discusses the use of these training sets to train several types of neural networks. Neural network configurations used include the adaptive resonance network, the back-propagation-trained network, and the counter-propagation network. This work shows that neural networks can be used to produce robust sequencers. These sequencers can learn by example to execute the step-by-step procedures of optical alignment and also can learn adaptively to correct for environmentally induced misalignment. The long-range objective is to use neural networks to automate the alignment and operation of optical measurement systems in remote, harsh, or dangerous aerospace environments. This work also shows that when neural networks are trained by a human operator, training sets should be recorded, training should be executed, and testing should be done in a manner that does not depend on intellectual judgments of the human operator.

Inversion of a lateral log using neural networks

International Nuclear Information System (INIS)

Garcia, G.; Whitman, W.W.

1992-01-01

In this paper a technique using neural networks is demonstrated for the inversion of a lateral log. The lateral log is simulated by a finite difference method which in turn is used as an input to a backpropagation neural network. An initial guess earth model is generated from the neural network, which is then input to a Marquardt inversion. The neural network reacts to gross and subtle data features in actual logs and produces a response inferred from the knowledge stored in the network during a training process. The neural network inversion of lateral logs is tested on synthetic and field data. Tests using field data resulted in a final earth model whose simulated lateral is in good agreement with the actual log data

Improved Extension Neural Network and Its Applications

Directory of Open Access Journals (Sweden)

Yu Zhou

2014-01-01

Full Text Available Extension neural network (ENN is a new neural network that is a combination of extension theory and artificial neural network (ANN. The learning algorithm of ENN is based on supervised learning algorithm. One of important issues in the field of classification and recognition of ENN is how to achieve the best possible classifier with a small number of labeled training data. Training data selection is an effective approach to solve this issue. In this work, in order to improve the supervised learning performance and expand the engineering application range of ENN, we use a novel data selection method based on shadowed sets to refine the training data set of ENN. Firstly, we use clustering algorithm to label the data and induce shadowed sets. Then, in the framework of shadowed sets, the samples located around each cluster centers (core data and the borders between clusters (boundary data are selected as training data. Lastly, we use selected data to train ENN. Compared with traditional ENN, the proposed improved ENN (IENN has a better performance. Moreover, IENN is independent of the supervised learning algorithms and initial labeled data. Experimental results verify the effectiveness and applicability of our proposed work.

Reconstruction of neutron spectra through neural networks

International Nuclear Information System (INIS)

Vega C, H.R.; Hernandez D, V.M.; Manzanares A, E.

2003-01-01

A neural network has been used to reconstruct the neutron spectra starting from the counting rates of the detectors of the Bonner sphere spectrophotometric system. A group of 56 neutron spectra was selected to calculate the counting rates that would produce in a Bonner sphere system, with these data and the spectra it was trained the neural network. To prove the performance of the net, 12 spectra were used, 6 were taken of the group used for the training, 3 were obtained of mathematical functions and those other 3 correspond to real spectra. When comparing the original spectra of those reconstructed by the net we find that our net has a poor performance when reconstructing monoenergetic spectra, this attributes it to those characteristic of the spectra used for the training of the neural network, however for the other groups of spectra the results of the net are appropriate with the prospective ones. (Author)

Training of reverse propagation neural networks applied to neutron dosimetry

International Nuclear Information System (INIS)

Hernandez P, C. F.; Martinez B, M. R.; Leon P, A. A.; Espinoza G, J. G.; Castaneda M, V. H.; Solis S, L. O.; Castaneda M, R.; Ortiz R, M.; Vega C, H. R.; Mendez V, R.; Gallego, E.; De Sousa L, M. A.

2016-10-01

Neutron dosimetry is of great importance in radiation protection as aims to provide dosimetric quantities to assess the magnitude of detrimental health effects due to exposure of neutron radiation. To quantify detriment to health is necessary to evaluate the dose received by the occupationally exposed personnel using different detection systems called dosimeters, which have very dependent responses to the energy distribution of neutrons. The neutron detection is a much more complex problem than the detection of charged particles, since it does not carry an electric charge, does not cause direct ionization and has a greater penetration power giving the possibility of interacting with matter in a different way. Because of this, various neutron detection systems have been developed, among which the Bonner spheres spectrometric system stands out due to the advantages that possesses, such as a wide range of energy, high sensitivity and easy operation. However, once obtained the counting rates, the problem lies in the neutron spectrum deconvolution, necessary for the calculation of the doses, using different mathematical methods such as Monte Carlo, maximum entropy, iterative methods among others, which present various difficulties that have motivated the development of new technologies. Nowadays, methods based on artificial intelligence technologies are being used to perform neutron dosimetry, mainly using the theory of artificial neural networks. In these new methods the need for spectrum reconstruction can be eliminated for the calculation of the doses. In this work an artificial neural network or reverse propagation was trained for the calculation of 15 equivalent doses from the counting rates of the Bonner spheres spectrometric system using a set of 7 spheres, one of 2 spheres and two of a single sphere of different sizes, testing different error values until finding the most appropriate. The optimum network topology was obtained through the robust design

Neural networks for sensor validation and plant monitoring

International Nuclear Information System (INIS)

Upadhyaya, B.R.; Eryurek, E.; Mathai, G.

1990-01-01

Sensor and process monitoring in power plants require the estimation of one or more process variables. Neural network paradigms are suitable for establishing general nonlinear relationships among a set of plant variables. Multiple-input multiple-output autoassociative networks can follow changes in plant-wide behavior. The backpropagation algorithm has been applied for training feedforward networks. A new and enhanced algorithm for training neural networks (BPN) has been developed and implemented in a VAX workstation. Operational data from the Experimental Breeder Reactor-II (EBR-II) have been used to study the performance of BPN. Several results of application to the EBR-II are presented

Logarithmic learning for generalized classifier neural network.

Science.gov (United States)

Ozyildirim, Buse Melis; Avci, Mutlu

2014-12-01

Generalized classifier neural network is introduced as an efficient classifier among the others. Unless the initial smoothing parameter value is close to the optimal one, generalized classifier neural network suffers from convergence problem and requires quite a long time to converge. In this work, to overcome this problem, a logarithmic learning approach is proposed. The proposed method uses logarithmic cost function instead of squared error. Minimization of this cost function reduces the number of iterations used for reaching the minima. The proposed method is tested on 15 different data sets and performance of logarithmic learning generalized classifier neural network is compared with that of standard one. Thanks to operation range of radial basis function included by generalized classifier neural network, proposed logarithmic approach and its derivative has continuous values. This makes it possible to adopt the advantage of logarithmic fast convergence by the proposed learning method. Due to fast convergence ability of logarithmic cost function, training time is maximally decreased to 99.2%. In addition to decrease in training time, classification performance may also be improved till 60%. According to the test results, while the proposed method provides a solution for time requirement problem of generalized classifier neural network, it may also improve the classification accuracy. The proposed method can be considered as an efficient way for reducing the time requirement problem of generalized classifier neural network. Copyright © 2014 Elsevier Ltd. All rights reserved.

A training rule which guarantees finite-region stability for a class of closed-loop neural-network control systems.

Science.gov (United States)

Kuntanapreeda, S; Fullmer, R R

1996-01-01

A training method for a class of neural network controllers is presented which guarantees closed-loop system stability. The controllers are assumed to be nonlinear, feedforward, sampled-data, full-state regulators implemented as single hidden-layer neural networks. The controlled systems must be locally hermitian and observable. Stability of the closed-loop system is demonstrated by determining a Lyapunov function, which can be used to identify a finite stability region about the regulator point.

Reducing Wind Tunnel Data Requirements Using Neural Networks

Science.gov (United States)

Ross, James C.; Jorgenson, Charles C.; Norgaard, Magnus

1997-01-01

The use of neural networks to minimize the amount of data required to completely define the aerodynamic performance of a wind tunnel model is examined. The accuracy requirements for commercial wind tunnel test data are very severe and are difficult to reproduce using neural networks. For the current work, multiple input, single output networks were trained using a Levenberg-Marquardt algorithm for each of the aerodynamic coefficients. When applied to the aerodynamics of a 55% scale model of a U.S. Air Force/ NASA generic fighter configuration, this scheme provided accurate models of the lift, drag, and pitching-moment coefficients. Using only 50% of the data acquired during, the wind tunnel test, the trained neural network had a predictive accuracy equal to or better than the accuracy of the experimental measurements.

Engine cylinder pressure reconstruction using crank kinematics and recurrently-trained neural networks

Science.gov (United States)

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.

Neural network application to diesel generator diagnostics

International Nuclear Information System (INIS)

Logan, K.P.

1990-01-01

Diagnostic problems typically begin with the observation of some system behavior which is recognized as a deviation from the expected. The fundamental underlying process is one involving pattern matching cf observed symptoms to a set of compiled symptoms belonging to a fault-symptom mapping. Pattern recognition is often relied upon for initial fault detection and diagnosis. Parallel distributed processing (PDP) models employing neural network paradigms are known to be good pattern recognition devices. This paper describes the application of neural network processing techniques to the malfunction diagnosis of subsystems within a typical diesel generator configuration. Neural network models employing backpropagation learning were developed to correctly recognize fault conditions from the input diagnostic symptom patterns pertaining to various engine subsystems. The resulting network models proved to be excellent pattern recognizers for malfunction examples within the training set. The motivation for employing network models in lieu of a rule-based expert system, however, is related to the network's potential for generalizing malfunctions outside of the training set, as in the case of noisy or partial symptom patterns

Prediction based chaos control via a new neural network

International Nuclear Information System (INIS)

Shen Liqun; Wang Mao; Liu Wanyu; Sun Guanghui

2008-01-01

In this Letter, a new chaos control scheme based on chaos prediction is proposed. To perform chaos prediction, a new neural network architecture for complex nonlinear approximation is proposed. And the difficulty in building and training the neural network is also reduced. Simulation results of Logistic map and Lorenz system show the effectiveness of the proposed chaos control scheme and the proposed neural network

Training and validation of the ATLAS pixel clustering neural networks

CERN Document Server

The ATLAS collaboration

2018-01-01

The high centre-of-mass energy of the LHC gives rise to dense environments, such as the core of high-pT jets, in which the charge clusters left by ionising particles in the silicon sensors of the pixel detector can merge, compromising the tracking and vertexing efficiency. To recover optimal performance, a neural network-based approach is used to separate clusters originating from single and multiple particles and to estimate all hit positions within clusters. This note presents the training strategy employed and a set of benchmark performance measurements on a Monte Carlo sample of high-pT dijet events.

International Conference on Artificial Neural Networks (ICANN)

CERN Document Server

Mladenov, Valeri; Kasabov, Nikola; Artificial Neural Networks : Methods and Applications in Bio-/Neuroinformatics

2015-01-01

The book reports on the latest theories on artificial neural networks, with a special emphasis on bio-neuroinformatics methods. It includes twenty-three papers selected from among the best contributions on bio-neuroinformatics-related issues, which were presented at the International Conference on Artificial Neural Networks, held in Sofia, Bulgaria, on September 10-13, 2013 (ICANN 2013). The book covers a broad range of topics concerning the theory and applications of artificial neural networks, including recurrent neural networks, super-Turing computation and reservoir computing, double-layer vector perceptrons, nonnegative matrix factorization, bio-inspired models of cell communities, Gestalt laws, embodied theory of language understanding, saccadic gaze shifts and memory formation, and new training algorithms for Deep Boltzmann Machines, as well as dynamic neural networks and kernel machines. It also reports on new approaches to reinforcement learning, optimal control of discrete time-delay systems, new al...

SOLAR PHOTOVOLTAIC OUTPUT POWER FORECASTING USING BACK PROPAGATION NEURAL NETWORK

Directory of Open Access Journals (Sweden)

B. Jency Paulin

2016-01-01

Full Text Available Solar Energy is an important renewable and unlimited source of energy. Solar photovoltaic power forecasting, is an estimation of the expected power production, that help the grid operators to better manage the electric balance between power demand and supply. Neural network is a computational model that can predict new outcomes from past trends. The artificial neural network is used for photovoltaic plant energy forecasting. The output power for solar photovoltaic cell is predicted on hourly basis. In historical dataset collection process, two dataset was collected and used for analysis. The dataset was provided with three independent attributes and one dependent attributes. The implementation of Artificial Neural Network structure is done by Multilayer Perceptron (MLP and training procedure for neural network is done by error Back Propagation (BP. In order to train and test the neural network, the datasets are divided in the ratio 70:30. The accuracy of prediction can be done by using various error measurement criteria and the performance of neural network is to be noted.

Livermore Big Artificial Neural Network Toolkit

Energy Technology Data Exchange (ETDEWEB)

2016-07-01

LBANN is a toolkit that is designed to train artificial neural networks efficiently on high performance computing architectures. It is optimized to take advantages of key High Performance Computing features to accelerate neural network training. Specifically it is optimized for low-latency, high bandwidth interconnects, node-local NVRAM, node-local GPU accelerators, and high bandwidth parallel file systems. It is built on top of the open source Elemental distributed-memory dense and spars-direct linear algebra and optimization library that is released under the BSD license. The algorithms contained within LBANN are drawn from the academic literature and implemented to work within a distributed-memory framework.

Anomaly detection in an automated safeguards system using neural networks

International Nuclear Information System (INIS)

Whiteson, R.; Howell, J.A.

1992-01-01

An automated safeguards system must be able to detect an anomalous event, identify the nature of the event, and recommend a corrective action. Neural networks represent a new way of thinking about basic computational mechanisms for intelligent information processing. In this paper, we discuss the issues involved in applying a neural network model to the first step of this process: anomaly detection in materials accounting systems. We extend our previous model to a 3-tank problem and compare different neural network architectures and algorithms. We evaluate the computational difficulties in training neural networks and explore how certain design principles affect the problems. The issues involved in building a neural network architecture include how the information flows, how the network is trained, how the neurons in a network are connected, how the neurons process information, and how the connections between neurons are modified. Our approach is based on the demonstrated ability of neural networks to model complex, nonlinear, real-time processes. By modeling the normal behavior of the processes, we can predict how a system should be behaving and, therefore, detect when an abnormality occurs

Artificial Neural Networks for Nonlinear Dynamic Response Simulation in Mechanical Systems

DEFF Research Database (Denmark)

Christiansen, Niels Hørbye; Høgsberg, Jan Becker; Winther, Ole

2011-01-01

It is shown how artificial neural networks can be trained to predict dynamic response of a simple nonlinear structure. Data generated using a nonlinear finite element model of a simplified wind turbine is used to train a one layer artificial neural network. When trained properly the network is ab...... to perform accurate response prediction much faster than the corresponding finite element model. Initial result indicate a reduction in cpu time by two orders of magnitude....

Seismic signal auto-detecing from different features by using Convolutional Neural Network

Science.gov (United States)

Huang, Y.; Zhou, Y.; Yue, H.; Zhou, S.

2017-12-01

We try Convolutional Neural Network to detect some features of seismic data and compare their efficience. The features include whether a signal is seismic signal or noise and the arrival time of P and S phase and each feature correspond to a Convolutional Neural Network. We first use traditional STA/LTA to recongnize some events and then use templete matching to find more events as training set for the Neural Network. To make the training set more various, we add some noise to the seismic data and make some synthetic seismic data and noise. The 3-component raw signal and time-frequancy ananlyze are used as the input data for our neural network. Our Training is performed on GPUs to achieve efficient convergence. Our method improved the precision in comparison with STA/LTA and template matching. We will move to recurrent neural network to see if this kind network is better in detect P and S phase.

Decoding small surface codes with feedforward neural networks

Science.gov (United States)

Varsamopoulos, Savvas; Criger, Ben; Bertels, Koen

2018-01-01

Surface codes reach high error thresholds when decoded with known algorithms, but the decoding time will likely exceed the available time budget, especially for near-term implementations. To decrease the decoding time, we reduce the decoding problem to a classification problem that a feedforward neural network can solve. We investigate quantum error correction and fault tolerance at small code distances using neural network-based decoders, demonstrating that the neural network can generalize to inputs that were not provided during training and that they can reach similar or better decoding performance compared to previous algorithms. We conclude by discussing the time required by a feedforward neural network decoder in hardware.

Neural Network Models for Free Radical Polymerization of Methyl Methacrylate

International Nuclear Information System (INIS)

Curteanu, S.; Leon, F.; Galea, D.

2003-01-01

In this paper, a neural network modeling of the batch bulk methyl methacrylate polymerization is performed. To obtain conversion, number and weight average molecular weights, three neural networks were built. Each was a multilayer perception with one or two hidden layers. The choice of network topology, i.e. the number of hidden layers and the number of neurons in these layers, was based on achieving a compromise between precision and complexity. Thus, it was intended to have an error as small as possible at the end of back-propagation training phases, while using a network with reduced complexity. The performances of the networks were evaluated by comparing network predictions with training data, validation data (which were not uses for training), and with the results of a mechanistic model. The accurate predictions of neural networks for monomer conversion, number average molecular weight and weight average molecular weight proves that this modeling methodology gives a good representation and generalization of the batch bulk methyl methacrylate polymerization. (author)

Iris double recognition based on modified evolutionary neural network

Science.gov (United States)

Liu, Shuai; Liu, Yuan-Ning; Zhu, Xiao-Dong; Huo, Guang; Liu, Wen-Tao; Feng, Jia-Kai

2017-11-01

Aiming at multicategory iris recognition under illumination and noise interference, this paper proposes a method of iris double recognition based on a modified evolutionary neural network. An equalization histogram and Laplace of Gaussian operator are used to process the iris to suppress illumination and noise interference and Haar wavelet to convert the iris feature to binary feature encoding. Calculate the Hamming distance for the test iris and template iris , and compare with classification threshold, determine the type of iris. If the iris cannot be identified as a different type, there needs to be a secondary recognition. The connection weights in back-propagation (BP) neural network use modified evolutionary neural network to adaptively train. The modified neural network is composed of particle swarm optimization with mutation operator and BP neural network. According to different iris libraries in different circumstances of experimental results, under illumination and noise interference, the correct recognition rate of this algorithm is higher, the ROC curve is closer to the coordinate axis, the training and recognition time is shorter, and the stability and the robustness are better.

Classification of conductance traces with recurrent neural networks

Science.gov (United States)

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.

DCS-Neural-Network Program for Aircraft Control and Testing

Science.gov (United States)

Jorgensen, Charles C.

2006-01-01

A computer program implements a dynamic-cell-structure (DCS) artificial neural network that can perform such tasks as learning selected aerodynamic characteristics of an airplane from wind-tunnel test data and computing real-time stability and control derivatives of the airplane for use in feedback linearized control. A DCS neural network is one of several types of neural networks that can incorporate additional nodes in order to rapidly learn increasingly complex relationships between inputs and outputs. In the DCS neural network implemented by the present program, the insertion of nodes is based on accumulated error. A competitive Hebbian learning rule (a supervised-learning rule in which connection weights are adjusted to minimize differences between actual and desired outputs for training examples) is used. A Kohonen-style learning rule (derived from a relatively simple training algorithm, implements a Delaunay triangulation layout of neurons) is used to adjust node positions during training. Neighborhood topology determines which nodes are used to estimate new values. The network learns, starting with two nodes, and adds new nodes sequentially in locations chosen to maximize reductions in global error. At any given time during learning, the error becomes homogeneously distributed over all nodes.

Convolutional neural networks based on augmented training samples for synthetic aperture radar target recognition

Science.gov (United States)

Yan, Yue

2018-03-01

A synthetic aperture radar (SAR) automatic target recognition (ATR) method based on the convolutional neural networks (CNN) trained by augmented training samples is proposed. To enhance the robustness of CNN to various extended operating conditions (EOCs), the original training images are used to generate the noisy samples at different signal-to-noise ratios (SNRs), multiresolution representations, and partially occluded images. Then, the generated images together with the original ones are used to train a designed CNN for target recognition. The augmented training samples can contrapuntally improve the robustness of the trained CNN to the covered EOCs, i.e., the noise corruption, resolution variance, and partial occlusion. Moreover, the significantly larger training set effectively enhances the representation capability for other conditions, e.g., the standard operating condition (SOC), as well as the stability of the network. Therefore, better performance can be achieved by the proposed method for SAR ATR. For experimental evaluation, extensive experiments are conducted on the Moving and Stationary Target Acquisition and Recognition dataset under SOC and several typical EOCs.

Probing many-body localization with neural networks

Science.gov (United States)

Schindler, Frank; Regnault, Nicolas; Neupert, Titus

2017-06-01

We show that a simple artificial neural network trained on entanglement spectra of individual states of a many-body quantum system can be used to determine the transition between a many-body localized and a thermalizing regime. Specifically, we study the Heisenberg spin-1/2 chain in a random external field. We employ a multilayer perceptron with a single hidden layer, which is trained on labeled entanglement spectra pertaining to the fully localized and fully thermal regimes. We then apply this network to classify spectra belonging to states in the transition region. For training, we use a cost function that contains, in addition to the usual error and regularization parts, a term that favors a confident classification of the transition region states. The resulting phase diagram is in good agreement with the one obtained by more conventional methods and can be computed for small systems. In particular, the neural network outperforms conventional methods in classifying individual eigenstates pertaining to a single disorder realization. It allows us to map out the structure of these eigenstates across the transition with spatial resolution. Furthermore, we analyze the network operation using the dreaming technique to show that the neural network correctly learns by itself the power-law structure of the entanglement spectra in the many-body localized regime.

Manifold absolute pressure estimation using neural network with hybrid training algorithm.

Directory of Open Access Journals (Sweden)

Mohd Taufiq Muslim

Full Text Available In a modern small gasoline engine fuel injection system, the load of the engine is estimated based on the measurement of the manifold absolute pressure (MAP sensor, which took place in the intake manifold. This paper present a more economical approach on estimating the MAP by using only the measurements of the throttle position and engine speed, resulting in lower implementation cost. The estimation was done via two-stage multilayer feed-forward neural network by combining Levenberg-Marquardt (LM algorithm, Bayesian Regularization (BR algorithm and Particle Swarm Optimization (PSO algorithm. Based on the results found in 20 runs, the second variant of the hybrid algorithm yields a better network performance than the first variant of hybrid algorithm, LM, LM with BR and PSO by estimating the MAP closely to the simulated MAP values. By using a valid experimental training data, the estimator network that trained with the second variant of the hybrid algorithm showed the best performance among other algorithms when used in an actual retrofit fuel injection system (RFIS. The performance of the estimator was also validated in steady-state and transient condition by showing a closer MAP estimation to the actual value.

Manifold absolute pressure estimation using neural network with hybrid training algorithm.

Science.gov (United States)

Muslim, Mohd Taufiq; Selamat, Hazlina; Alimin, Ahmad Jais; Haniff, Mohamad Fadzli

2017-01-01

In a modern small gasoline engine fuel injection system, the load of the engine is estimated based on the measurement of the manifold absolute pressure (MAP) sensor, which took place in the intake manifold. This paper present a more economical approach on estimating the MAP by using only the measurements of the throttle position and engine speed, resulting in lower implementation cost. The estimation was done via two-stage multilayer feed-forward neural network by combining Levenberg-Marquardt (LM) algorithm, Bayesian Regularization (BR) algorithm and Particle Swarm Optimization (PSO) algorithm. Based on the results found in 20 runs, the second variant of the hybrid algorithm yields a better network performance than the first variant of hybrid algorithm, LM, LM with BR and PSO by estimating the MAP closely to the simulated MAP values. By using a valid experimental training data, the estimator network that trained with the second variant of the hybrid algorithm showed the best performance among other algorithms when used in an actual retrofit fuel injection system (RFIS). The performance of the estimator was also validated in steady-state and transient condition by showing a closer MAP estimation to the actual value.

Bio-inspired spiking neural network for nonlinear systems control.

Science.gov (United States)

Pérez, Javier; Cabrera, Juan A; Castillo, Juan J; Velasco, Juan M

2018-08-01

Spiking neural networks (SNN) are the third generation of artificial neural networks. SNN are the closest approximation to biological neural networks. SNNs make use of temporal spike trains to command inputs and outputs, allowing a faster and more complex computation. As demonstrated by biological organisms, they are a potentially good approach to designing controllers for highly nonlinear dynamic systems in which the performance of controllers developed by conventional techniques is not satisfactory or difficult to implement. SNN-based controllers exploit their ability for online learning and self-adaptation to evolve when transferred from simulations to the real world. SNN's inherent binary and temporary way of information codification facilitates their hardware implementation compared to analog neurons. Biological neural networks often require a lower number of neurons compared to other controllers based on artificial neural networks. In this work, these neuronal systems are imitated to perform the control of non-linear dynamic systems. For this purpose, a control structure based on spiking neural networks has been designed. Particular attention has been paid to optimizing the structure and size of the neural network. The proposed structure is able to control dynamic systems with a reduced number of neurons and connections. A supervised learning process using evolutionary algorithms has been carried out to perform controller training. The efficiency of the proposed network has been verified in two examples of dynamic systems control. Simulations show that the proposed control based on SNN exhibits superior performance compared to other approaches based on Neural Networks and SNNs. Copyright © 2018 Elsevier Ltd. All rights reserved.

Intelligent neural network diagnostic system

International Nuclear Information System (INIS)

Mohamed, A.H.

2010-01-01

Recently, artificial neural network (ANN) has made a significant mark in the domain of diagnostic applications. Neural networks are used to implement complex non-linear mappings (functions) using simple elementary units interrelated through connections with adaptive weights. The performance of the ANN is mainly depending on their topology structure and weights. Some systems have been developed using genetic algorithm (GA) to optimize the topology of the ANN. But, they suffer from some limitations. They are : (1) The computation time requires for training the ANN several time reaching for the average weight required, (2) Slowness of GA for optimization process and (3) Fitness noise appeared in the optimization of ANN. This research suggests new issues to overcome these limitations for finding optimal neural network architectures to learn particular problems. This proposed methodology is used to develop a diagnostic neural network system. It has been applied for a 600 MW turbo-generator as a case of real complex systems. The proposed system has proved its significant performance compared to two common methods used in the diagnostic applications.

Parameter estimation in space systems using recurrent neural networks

Science.gov (United States)

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.

Influence of the Training Set Value on the Quality of the Neural Network to Identify Selected Moulding Sand Properties

Directory of Open Access Journals (Sweden)

Jakubski J.

2013-06-01

Full Text Available Artificial neural networks are one of the modern methods of the production optimisation. An attempt to apply neural networks for controlling the quality of bentonite moulding sands is presented in this paper. This is the assessment method of sands suitability by means of detecting correlations between their individual parameters. This paper presents the next part of the study on usefulness of artificial neural networks to support rebonding of green moulding sand, using chosen properties of moulding sands, which can be determined fast. The effect of changes in the training set quantity on the quality of the network is presented in this article. It has been shown that a small change in the data set would change the quality of the network, and may also make it necessary to change the type of network in order to obtain good results.

Neural network error correction for solving coupled ordinary differential equations

Science.gov (United States)

Shelton, R. O.; Darsey, J. A.; Sumpter, B. G.; Noid, D. W.

1992-01-01

A neural network is presented to learn errors generated by a numerical algorithm for solving coupled nonlinear differential equations. The method is based on using a neural network to correctly learn the error generated by, for example, Runge-Kutta on a model molecular dynamics (MD) problem. The neural network programs used in this study were developed by NASA. Comparisons are made for training the neural network using backpropagation and a new method which was found to converge with fewer iterations. The neural net programs, the MD model and the calculations are discussed.

Electricity price forecast using Combinatorial Neural Network trained by a new stochastic search method

International Nuclear Information System (INIS)

Abedinia, O.; Amjady, N.; Shafie-khah, M.; Catalão, J.P.S.

2015-01-01

Highlights: • Presenting a Combinatorial Neural Network. • Suggesting a new stochastic search method. • Adapting the suggested method as a training mechanism. • Proposing a new forecast strategy. • Testing the proposed strategy on real-world electricity markets. - Abstract: Electricity price forecast is key information for successful operation of electricity market participants. However, the time series of electricity price has nonlinear, non-stationary and volatile behaviour and so its forecast method should have high learning capability to extract the complex input/output mapping function of electricity price. In this paper, a Combinatorial Neural Network (CNN) based forecasting engine is proposed to predict the future values of price data. The CNN-based forecasting engine is equipped with a new training mechanism for optimizing the weights of the CNN. This training mechanism is based on an efficient stochastic search method, which is a modified version of chemical reaction optimization algorithm, giving high learning ability to the CNN. The proposed price forecast strategy is tested on the real-world electricity markets of Pennsylvania–New Jersey–Maryland (PJM) and mainland Spain and its obtained results are extensively compared with the results obtained from several other forecast methods. These comparisons illustrate effectiveness of the proposed strategy.

Radial basis function neural network for power system load-flow

International Nuclear Information System (INIS)

Karami, A.; Mohammadi, M.S.

2008-01-01

This paper presents a method for solving the load-flow problem of the electric power systems using radial basis function (RBF) neural network with a fast hybrid training method. The main idea is that some operating conditions (values) are needed to solve the set of non-linear algebraic equations of load-flow by employing an iterative numerical technique. Therefore, we may view the outputs of a load-flow program as functions of the operating conditions. Indeed, we are faced with a function approximation problem and this can be done by an RBF neural network. The proposed approach has been successfully applied to the 10-machine and 39-bus New England test system. In addition, this method has been compared with that of a multi-layer perceptron (MLP) neural network model. The simulation results show that the RBF neural network is a simpler method to implement and requires less training time to converge than the MLP neural network. (author)

A pre-trained convolutional neural network based method for thyroid nodule diagnosis.

Science.gov (United States)

Ma, Jinlian; Wu, Fa; Zhu, Jiang; Xu, Dong; Kong, Dexing

2017-01-01

In ultrasound images, most thyroid nodules are in heterogeneous appearances with various internal components and also have vague boundaries, so it is difficult for physicians to discriminate malignant thyroid nodules from benign ones. In this study, we propose a hybrid method for thyroid nodule diagnosis, which is a fusion of two pre-trained convolutional neural networks (CNNs) with different convolutional layers and fully-connected layers. Firstly, the two networks pre-trained with ImageNet database are separately trained. Secondly, we fuse feature maps learned by trained convolutional filters, pooling and normalization operations of the two CNNs. Finally, with the fused feature maps, a softmax classifier is used to diagnose thyroid nodules. The proposed method is validated on 15,000 ultrasound images collected from two local hospitals. Experiment results show that the proposed CNN based methods can accurately and effectively diagnose thyroid nodules. In addition, the fusion of the two CNN based models lead to significant performance improvement, with an accuracy of 83.02%±0.72%. These demonstrate the potential clinical applications of this method. Copyright © 2016 Elsevier B.V. All rights reserved.

Computer interpretation of thallium SPECT studies based on neural network analysis

Science.gov (United States)

Wang, David C.; Karvelis, K. C.

1991-06-01

A class of artificial intelligence (Al) programs known as neural networks are well suited to pattern recognition. A neural network is trained rather than programmed to recognize patterns. This differs from "expert system" Al programs in that it is not following an extensive set of rules determined by the programmer, but rather bases its decision on a gestalt interpretation of the image. The "bullseye" images from cardiac stress thallium tests performed on 50 male patients, as well as several simulated images were used to train the network. The network was able to accurately classify all patients in the training set. The network was then tested against 50 unknown patients and was able to correctly categorize 77% of the areas of ischemia and 92% of the areas of infarction. While not yet matching the ability of a trained physician, the neural network shows great promise in this area and has potential application in other areas of medical imaging.

Computer interpretation of thallium SPECT studies based on neural network analysis

International Nuclear Information System (INIS)

Wang, D.C.; Karvelis, K.C.

1991-01-01

This paper reports that a class of artificial intelligence (AI) programs known as neural-networks are well suited to pattern recognition. A neural network is trained rather than programmed to recognize patterns. This differs from expert system AI programs in that it is not following an extensive set of rules determined by the programmer, but rather bases its decision on a gestalt interpretation of the image. The bullseye images from cardiac stress thallium tests performed on 50 male patients, as well as several simulated images were used to train the network. The network was able to accurately classify all patients in the training set. The network was then tested against 50 unknown patients and was able to correctly categorize 77% of the areas of ischemia and 92% of the areas of infarction. While not yet matching the ability of the trained physician, the neural network shows great promise in this area and has potential application in other areas of medical imaging

A fuzzy neural network for sensor signal estimation

International Nuclear Information System (INIS)

Na, Man Gyun

2000-01-01

In this work, a fuzzy neural network is used to estimate the relevant sensor signal using other sensor signals. Noise components in input signals into the fuzzy neural network are removed through the wavelet denoising technique. Principal component analysis (PCA) is used to reduce the dimension of an input space without losing a significant amount of information. A lower dimensional input space will also usually reduce the time necessary to train a fuzzy-neural network. Also, the principal component analysis makes easy the selection of the input signals into the fuzzy neural network. The fuzzy neural network parameters are optimized by two learning methods. A genetic algorithm is used to optimize the antecedent parameters of the fuzzy neural network and a least-squares algorithm is used to solve the consequent parameters. The proposed algorithm was verified through the application to the pressurizer water level and the hot-leg flowrate measurements in pressurized water reactors

An artifical neural network for detection of simulated dental caries

Energy Technology Data Exchange (ETDEWEB)

Kositbowornchai, S. [Khon Kaen Univ. (Thailand). Dept. of Oral Diagnosis; Siriteptawee, S.; Plermkamon, S.; Bureerat, S. [Khon Kaen Univ. (Thailand). Dept. of Mechanical Engineering; Chetchotsak, D. [Khon Kaen Univ. (Thailand). Dept. of Industrial Engineering

2006-08-15

Objects: A neural network was developed to diagnose artificial dental caries using images from a charged-coupled device (CCD)camera and intra-oral digital radiography. The diagnostic performance of this neural network was evaluated against a gold standard. Materials and methods: The neural network design was the Learning Vector Quantization (LVQ) used to classify a tooth surface as sound or as having dental caries. The depth of the dental caries was indicated on a graphic user interface (GUI) screen developed by Matlab programming. Forty-nine images of both sound and simulated dental caries, derived from a CCD camera and by digital radiography, were used to 'train' an artificial neural network. After the 'training' process, a separate test-set comprising 322 unseen images was evaluated. Tooth sections and microscopic examinations were used to confirm the actual dental caries status.The performance of neural network was evaluated using diagnostic test. Results: The sensitivity (95%CI)/specificity (95%CI) of dental caries detection by the CCD camera and digital radiography were 0.77(0.68-0.85)/0.85(0.75-0.92) and 0.81(0.72-0.88)/0.93(0.84-0.97), respectively. The accuracy of caries depth-detection by the CCD camera and digital radiography was 58 and 40%, respectively. Conclusions: The model neural network used in this study could be a prototype for caries detection but should be improved for classifying caries depth. Our study suggests an artificial neural network can be trained to make the correct interpretations of dental caries. (orig.)

An artifical neural network for detection of simulated dental caries

International Nuclear Information System (INIS)

Kositbowornchai, S.; Siriteptawee, S.; Plermkamon, S.; Bureerat, S.; Chetchotsak, D.

2006-01-01

Objects: A neural network was developed to diagnose artificial dental caries using images from a charged-coupled device (CCD)camera and intra-oral digital radiography. The diagnostic performance of this neural network was evaluated against a gold standard. Materials and methods: The neural network design was the Learning Vector Quantization (LVQ) used to classify a tooth surface as sound or as having dental caries. The depth of the dental caries was indicated on a graphic user interface (GUI) screen developed by Matlab programming. Forty-nine images of both sound and simulated dental caries, derived from a CCD camera and by digital radiography, were used to 'train' an artificial neural network. After the 'training' process, a separate test-set comprising 322 unseen images was evaluated. Tooth sections and microscopic examinations were used to confirm the actual dental caries status.The performance of neural network was evaluated using diagnostic test. Results: The sensitivity (95%CI)/specificity (95%CI) of dental caries detection by the CCD camera and digital radiography were 0.77(0.68-0.85)/0.85(0.75-0.92) and 0.81(0.72-0.88)/0.93(0.84-0.97), respectively. The accuracy of caries depth-detection by the CCD camera and digital radiography was 58 and 40%, respectively. Conclusions: The model neural network used in this study could be a prototype for caries detection but should be improved for classifying caries depth. Our study suggests an artificial neural network can be trained to make the correct interpretations of dental caries. (orig.)

Design of Neural Networks for Fast Convergence and Accuracy: Dynamics and Control

Science.gov (United States)

Maghami, Peiman G.; Sparks, Dean W., Jr.

1997-01-01

A procedure for the design and training of artificial neural networks, used for rapid and efficient controls and dynamics design and analysis for flexible space systems, has been developed. Artificial neural networks are employed, such that once properly trained, they provide a means of evaluating the impact of design changes rapidly. Specifically, two-layer feedforward neural networks are designed to approximate the functional relationship between the component/spacecraft design changes and measures of its performance or nonlinear dynamics of the system/components. A training algorithm, based on statistical sampling theory, is presented, which guarantees that the trained networks provide a designer-specified degree of accuracy in mapping the functional relationship. Within each iteration of this statistical-based algorithm, a sequential design algorithm is used for the design and training of the feedforward network to provide rapid convergence to the network goals. Here, at each sequence a new network is trained to minimize the error of previous network. The proposed method should work for applications wherein an arbitrary large source of training data can be generated. Two numerical examples are performed on a spacecraft application in order to demonstrate the feasibility of the proposed approach.

Control of autonomous robot using neural networks

Science.gov (United States)

Barton, Adam; Volna, Eva

2017-07-01

The aim of the article is to design a method of control of an autonomous robot using artificial neural networks. The introductory part describes control issues from the perspective of autonomous robot navigation and the current mobile robots controlled by neural networks. The core of the article is the design of the controlling neural network, and generation and filtration of the training set using ART1 (Adaptive Resonance Theory). The outcome of the practical part is an assembled Lego Mindstorms EV3 robot solving the problem of avoiding obstacles in space. To verify models of an autonomous robot behavior, a set of experiments was created as well as evaluation criteria. The speed of each motor was adjusted by the controlling neural network with respect to the situation in which the robot was found.

Learning Data Set Influence on Identification Accuracy of Gas Turbine Neural Network Model

Science.gov (United States)

Kuznetsov, A. V.; Makaryants, G. M.

2018-01-01

There are many gas turbine engine identification researches via dynamic neural network models. It should minimize errors between model and real object during identification process. Questions about training data set processing of neural networks are usually missed. This article presents a study about influence of data set type on gas turbine neural network model accuracy. The identification object is thermodynamic model of micro gas turbine engine. The thermodynamic model input signal is the fuel consumption and output signal is the engine rotor rotation frequency. Four types input signals was used for creating training and testing data sets of dynamic neural network models - step, fast, slow and mixed. Four dynamic neural networks were created based on these types of training data sets. Each neural network was tested via four types test data sets. In the result 16 transition processes from four neural networks and four test data sets from analogous solving results of thermodynamic model were compared. The errors comparison was made between all neural network errors in each test data set. In the comparison result it was shown error value ranges of each test data set. It is shown that error values ranges is small therefore the influence of data set types on identification accuracy is low.

UNMANNED AIR VEHICLE STABILIZATION BASED ON NEURAL NETWORK REGULATOR

Directory of Open Access Journals (Sweden)

S. S. Andropov

2016-09-01

Full Text Available A problem of stabilizing for the multirotor unmanned aerial vehicle in an environment with external disturbances is researched. A classic proportional-integral-derivative controller is analyzed, its flaws are outlined: inability to respond to changing of external conditions and the need for manual adjustment of coefficients. The paper presents an adaptive adjustment method for coefficients of the proportional-integral-derivative controller based on neural networks. A neural network structure, its input and output data are described. Neural networks with three layers are used to create an adaptive stabilization system for the multirotor unmanned aerial vehicle. Training of the networks is done with the back propagation method. Each neural network produces regulator coefficients for each angle of stabilization as its output. A method for network training is explained. Several graphs of transition process on different stages of learning, including processes with external disturbances, are presented. It is shown that the system meets stabilization requirements with sufficient number of iterations. Described adjustment method for coefficients can be used in remote control of unmanned aerial vehicles, operating in the changing environment.

Empirical modeling of nuclear power plants using neural networks

International Nuclear Information System (INIS)

Parlos, A.G.; Atiya, A.; Chong, K.T.

1991-01-01

A summary of a procedure for nonlinear identification of process dynamics encountered in nuclear power plant components is presented in this paper using artificial neural systems. A hybrid feedforward/feedback neural network, namely, a recurrent multilayer perceptron, is used as the nonlinear structure for system identification. In the overall identification process, 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 time-dependent system nonlinearities. The standard backpropagation learning algorithm is modified and is used to train the proposed hybrid network in a supervised manner. The performance of recurrent multilayer perceptron networks in identifying process dynamics is investigated via the case study of a U-tube steam generator. The nonlinear response of a representative steam generator is predicted using a neural network and is compared to the response obtained from a sophisticated physical model during both high- and low-power operation. The transient responses compare well, though further research is warranted for training and testing of recurrent neural networks during more severe operational transients and accident scenarios

Feed forward neural networks modeling for K-P interactions

International Nuclear Information System (INIS)

El-Bakry, M.Y.

2003-01-01

Artificial intelligence techniques involving neural networks became vital modeling tools where model dynamics are difficult to track with conventional techniques. The paper make use of the feed forward neural networks (FFNN) to model the charged multiplicity distribution of K-P interactions at high energies. The FFNN was trained using experimental data for the multiplicity distributions at different lab momenta. Results of the FFNN model were compared to that generated using the parton two fireball model and the experimental data. The proposed FFNN model results showed good fitting to the experimental data. The neural network model performance was also tested at non-trained space and was found to be in good agreement with the experimental data

Fast, Simple and Accurate Handwritten Digit Classification by Training Shallow Neural Network Classifiers with the 'Extreme Learning Machine' Algorithm.

Directory of Open Access Journals (Sweden)

Mark D McDonnell

Full Text Available Recent advances in training deep (multi-layer architectures have inspired a renaissance in neural network use. For example, deep convolutional networks are becoming the default option for difficult tasks on large datasets, such as image and speech recognition. However, here we show that error rates below 1% on the MNIST handwritten digit benchmark can be replicated with shallow non-convolutional neural networks. This is achieved by training such networks using the 'Extreme Learning Machine' (ELM approach, which also enables a very rapid training time (∼ 10 minutes. Adding distortions, as is common practise for MNIST, reduces error rates even further. Our methods are also shown to be capable of achieving less than 5.5% error rates on the NORB image database. To achieve these results, we introduce several enhancements to the standard ELM algorithm, which individually and in combination can significantly improve performance. The main innovation is to ensure each hidden-unit operates only on a randomly sized and positioned patch of each image. This form of random 'receptive field' sampling of the input ensures the input weight matrix is sparse, with about 90% of weights equal to zero. Furthermore, combining our methods with a small number of iterations of a single-batch backpropagation method can significantly reduce the number of hidden-units required to achieve a particular performance. Our close to state-of-the-art results for MNIST and NORB suggest that the ease of use and accuracy of the ELM algorithm for designing a single-hidden-layer neural network classifier should cause it to be given greater consideration either as a standalone method for simpler problems, or as the final classification stage in deep neural networks applied to more difficult problems.

Higher-order neural network software for distortion invariant object recognition

Science.gov (United States)

Reid, Max B.; Spirkovska, Lilly

1991-01-01

The state-of-the-art in pattern recognition for such applications as automatic target recognition and industrial robotic vision relies on digital image processing. We present a higher-order neural network model and software which performs the complete feature extraction-pattern classification paradigm required for automatic pattern recognition. Using a third-order neural network, we demonstrate complete, 100 percent accurate invariance to distortions of scale, position, and in-plate rotation. In a higher-order neural network, feature extraction is built into the network, and does not have to be learned. Only the relatively simple classification step must be learned. This is key to achieving very rapid training. The training set is much smaller than with standard neural network software because the higher-order network only has to be shown one view of each object to be learned, not every possible view. The software and graphical user interface run on any Sun workstation. Results of the use of the neural software in autonomous robotic vision systems are presented. Such a system could have extensive application in robotic manufacturing.

Classification of E-Nose Aroma Data of Four Fruit Types by ABC-Based Neural Network

Directory of Open Access Journals (Sweden)

M. Fatih Adak

2016-02-01

Full Text Available Electronic nose technology is used in many areas, and frequently in the beverage industry for classification and quality-control purposes. In this study, four different aroma data (strawberry, lemon, cherry, and melon were obtained using a MOSES II electronic nose for the purpose of fruit classification. To improve the performance of the classification, the training phase of the neural network with two hidden layers was optimized using artificial bee colony algorithm (ABC, which is known to be successful in exploration. Test data were given to two different neural networks, each of which were trained separately with backpropagation (BP and ABC, and average test performances were measured as 60% for the artificial neural network trained with BP and 76.39% for the artificial neural network trained with ABC. Training and test phases were repeated 30 times to obtain these average performance measurements. This level of performance shows that the artificial neural network trained with ABC is successful in classifying aroma data.

Classification of E-Nose Aroma Data of Four Fruit Types by ABC-Based Neural Network.

Science.gov (United States)

Adak, M Fatih; Yumusak, Nejat

2016-02-27

Electronic nose technology is used in many areas, and frequently in the beverage industry for classification and quality-control purposes. In this study, four different aroma data (strawberry, lemon, cherry, and melon) were obtained using a MOSES II electronic nose for the purpose of fruit classification. To improve the performance of the classification, the training phase of the neural network with two hidden layers was optimized using artificial bee colony algorithm (ABC), which is known to be successful in exploration. Test data were given to two different neural networks, each of which were trained separately with backpropagation (BP) and ABC, and average test performances were measured as 60% for the artificial neural network trained with BP and 76.39% for the artificial neural network trained with ABC. Training and test phases were repeated 30 times to obtain these average performance measurements. This level of performance shows that the artificial neural network trained with ABC is successful in classifying aroma data.

Decorrelated Jet Substructure Tagging using Adversarial Neural Networks

CERN Multimedia

CERN. Geneva

2017-01-01

We describe a strategy for constructing a neural network jet substructure tagger which powerfully discriminates boosted decay signals while remaining largely uncorrelated with the jet mass. This reduces the impact of systematic uncertainties in background modeling while enhancing signal purity, resulting in improved discovery significance relative to existing taggers. The network is trained using an adversarial strategy, resulting in a tagger that learns to balance classification accuracy with decorrelation. As a benchmark scenario, we consider the case where large-radius jets originating from a boosted Z' decay are discriminated from a background of nonresonant quark and gluon jets. We show that in the presence of systematic uncertainties on the background rate, our adversarially-trained, decorrelated tagger considerably outperforms a conventionally trained neural network, despite having a slightly worse signal-background separation power. We generalize the adversarial training technique to include a paramet...

Transient analysis for PWR reactor core using neural networks predictors

International Nuclear Information System (INIS)

Gueray, B.S.

2001-01-01

In this study, transient analysis for a Pressurized Water Reactor core has been performed. A lumped parameter approximation is preferred for that purpose, to describe the reactor core together with mechanism which play an important role in dynamic analysis. The dynamic behavior of the reactor core during transients is analyzed considering the transient initiating events, wich are an essential part of Safety Analysis Reports. several transients are simulated based on the employed core model. Simulation results are in accord the physical expectations. A neural network is developed to predict the future response of the reactor core, in advance. The neural network is trained using the simulation results of a number of representative transients. Structure of the neural network is optimized by proper selection of transfer functions for the neurons. Trained neural network is used to predict the future responses following an early observation of the changes in system variables. Estimated behaviour using the neural network is in good agreement with the simulation results for various for types of transients. Results of this study indicate that the designed neural network can be used as an estimator of the time dependent behavior of the reactor core under transient conditions

Artificial neural network intelligent method for prediction

Science.gov (United States)

Trifonov, Roumen; Yoshinov, Radoslav; Pavlova, Galya; Tsochev, Georgi

2017-09-01

Accounting and financial classification and prediction problems are high challenge and researchers use different methods to solve them. Methods and instruments for short time prediction of financial operations using artificial neural network are considered. The methods, used for prediction of financial data as well as the developed forecasting system with neural network are described in the paper. The architecture of a neural network used four different technical indicators, which are based on the raw data and the current day of the week is presented. The network developed is used for forecasting movement of stock prices one day ahead and consists of an input layer, one hidden layer and an output layer. The training method is algorithm with back propagation of the error. The main advantage of the developed system is self-determination of the optimal topology of neural network, due to which it becomes flexible and more precise The proposed system with neural network is universal and can be applied to various financial instruments using only basic technical indicators as input data.

SCYNet. Testing supersymmetric models at the LHC with neural networks

Energy Technology Data Exchange (ETDEWEB)

Bechtle, Philip; Belkner, Sebastian; Hamer, Matthias [Universitaet Bonn, Bonn (Germany); Dercks, Daniel [Universitaet Hamburg, Hamburg (Germany); Keller, Tim; Kraemer, Michael; Sarrazin, Bjoern; Schuette-Engel, Jan; Tattersall, Jamie [RWTH Aachen University, Institute for Theoretical Particle Physics and Cosmology, Aachen (Germany)

2017-10-15

SCYNet (SUSY Calculating Yield Net) is a tool for testing supersymmetric models against LHC data. It uses neural network regression for a fast evaluation of the profile likelihood ratio. Two neural network approaches have been developed: one network has been trained using the parameters of the 11-dimensional phenomenological Minimal Supersymmetric Standard Model (pMSSM-11) as an input and evaluates the corresponding profile likelihood ratio within milliseconds. It can thus be used in global pMSSM-11 fits without time penalty. In the second approach, the neural network has been trained using model-independent signature-related objects, such as energies and particle multiplicities, which were estimated from the parameters of a given new physics model. (orig.)

Modeling of quasistatic magnetic hysteresis with feed-forward neural networks

International Nuclear Information System (INIS)

Makaveev, Dimitre; Dupre, Luc; De Wulf, Marc; Melkebeek, Jan

2001-01-01

A modeling technique for rate-independent (quasistatic) scalar magnetic hysteresis is presented, using neural networks. Based on the theory of dynamic systems and the wiping-out and congruency properties of the classical scalar Preisach hysteresis model, the choice of a feed-forward neural network model is motivated. The neural network input parameters at each time step are the corresponding magnetic field strength and memory state, thereby assuring accurate prediction of the change of magnetic induction. For rate-independent hysteresis, the current memory state can be determined by the last extreme magnetic field strength and induction values, kept in memory. The choice of a network training set is motivated and the performance of the network is illustrated for a test set not used during training. Very accurate prediction of both major and minor hysteresis loops is observed, proving that the neural network technique is suitable for hysteresis modeling. [copyright] 2001 American Institute of Physics

Modeling and control of magnetorheological fluid dampers using neural networks

Science.gov (United States)

Wang, D. H.; Liao, W. H.

2005-02-01

Due to the inherent nonlinear nature of magnetorheological (MR) fluid dampers, one of the challenging aspects for utilizing these devices to achieve high system performance is the development of accurate models and control algorithms that can take advantage of their unique characteristics. In this paper, the direct identification and inverse dynamic modeling for MR fluid dampers using feedforward and recurrent neural networks are studied. The trained direct identification neural network model can be used to predict the damping force of the MR fluid damper on line, on the basis of the dynamic responses across the MR fluid damper and the command voltage, and the inverse dynamic neural network model can be used to generate the command voltage according to the desired damping force through supervised learning. The architectures and the learning methods of the dynamic neural network models and inverse neural network models for MR fluid dampers are presented, and some simulation results are discussed. Finally, the trained neural network models are applied to predict and control the damping force of the MR fluid damper. Moreover, validation methods for the neural network models developed are proposed and used to evaluate their performance. Validation results with different data sets indicate that the proposed direct identification dynamic model using the recurrent neural network can be used to predict the damping force accurately and the inverse identification dynamic model using the recurrent neural network can act as a damper controller to generate the command voltage when the MR fluid damper is used in a semi-active mode.

Identification of Abnormal System Noise Temperature Patterns in Deep Space Network Antennas Using Neural Network Trained Fuzzy Logic

Science.gov (United States)

Lu, Thomas; Pham, Timothy; Liao, Jason

2011-01-01

This paper presents the development of a fuzzy logic function trained by an artificial neural network to classify the system noise temperature (SNT) of antennas in the NASA Deep Space Network (DSN). The SNT data were classified into normal, marginal, and abnormal classes. The irregular SNT pattern was further correlated with link margin and weather data. A reasonably good correlation is detected among high SNT, low link margin and the effect of bad weather; however we also saw some unexpected non-correlations which merit further study in the future.

Deep Recurrent Convolutional Neural Network: Improving Performance For Speech Recognition

OpenAIRE

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...

Quantitative analysis of volatile organic compounds using ion mobility spectra and cascade correlation neural networks

Science.gov (United States)

Harrington, Peter DEB.; Zheng, Peng

1995-01-01

Ion Mobility Spectrometry (IMS) is a powerful technique for trace organic analysis in the gas phase. Quantitative measurements are difficult, because IMS has a limited linear range. Factors that may affect the instrument response are pressure, temperature, and humidity. Nonlinear calibration methods, such as neural networks, may be ideally suited for IMS. Neural networks have the capability of modeling complex systems. Many neural networks suffer from long training times and overfitting. Cascade correlation neural networks train at very fast rates. They also build their own topology, that is a number of layers and number of units in each layer. By controlling the decay parameter in training neural networks, reproducible and general models may be obtained.

Application of artificial neural network for medical image recognition and diagnostic decision making

International Nuclear Information System (INIS)

Asada, N.; Eiho, S.; Doi, K.; MacMahon, H.; Montner, S.M.; Giger, M.L.

1989-01-01

An artificial neural network has been applied for pattern recognition and used as a tool in an expert system. The purpose of this study is to examine the potential usefulness of the neural network approach in medical applications for image recognition and decision making. The authors designed multilayer feedforward neural networks with a back-propagation algorithm for our study. Using first-pass radionuclide ventriculograms, we attempted to identify the right and left ventricles of the heart and the lungs by training the neural network from patterns of time-activity curves. In a preliminary study, the neural network enabled identification of the lungs and heart chambers once the network was trained sufficiently by means of repeated entries of data from the same case

Quantum neural networks: Current status and prospects for development

Science.gov (United States)

Altaisky, M. V.; Kaputkina, N. E.; Krylov, V. A.

2014-11-01

The idea of quantum artificial neural networks, first formulated in [34], unites the artificial neural network concept with the quantum computation paradigm. Quantum artificial neural networks were first systematically considered in the PhD thesis by T. Menneer (1998). Based on the works of Menneer and Narayanan [42, 43], Kouda, Matsui, and Nishimura [35, 36], Altaisky [2, 68], Zhou [67], and others, quantum-inspired learning algorithms for neural networks were developed, and are now used in various training programs and computer games [29, 30]. The first practically realizable scaled hardware-implemented model of the quantum artificial neural network is obtained by D-Wave Systems, Inc. [33]. It is a quantum Hopfield network implemented on the basis of superconducting quantum interference devices (SQUIDs). In this work we analyze possibilities and underlying principles of an alternative way to implement quantum neural networks on the basis of quantum dots. A possibility of using quantum neural network algorithms in automated control systems, associative memory devices, and in modeling biological and social networks is examined.

Neural Network Machine Learning and Dimension Reduction for Data Visualization

Science.gov (United States)

Liles, Charles A.

2014-01-01

Neural network machine learning in computer science is a continuously developing field of study. Although neural network models have been developed which can accurately predict a numeric value or nominal classification, a general purpose method for constructing neural network architecture has yet to be developed. Computer scientists are often forced to rely on a trial-and-error process of developing and improving accurate neural network models. In many cases, models are constructed from a large number of input parameters. Understanding which input parameters have the greatest impact on the prediction of the model is often difficult to surmise, especially when the number of input variables is very high. This challenge is often labeled the "curse of dimensionality" in scientific fields. However, techniques exist for reducing the dimensionality of problems to just two dimensions. Once a problem's dimensions have been mapped to two dimensions, it can be easily plotted and understood by humans. The ability to visualize a multi-dimensional dataset can provide a means of identifying which input variables have the highest effect on determining a nominal or numeric output. Identifying these variables can provide a better means of training neural network models; models can be more easily and quickly trained using only input variables which appear to affect the outcome variable. The purpose of this project is to explore varying means of training neural networks and to utilize dimensional reduction for visualizing and understanding complex datasets.

Learning drifting concepts with neural networks

NARCIS (Netherlands)

Biehl, Michael; Schwarze, Holm

1993-01-01

The learning of time-dependent concepts with a neural network is studied analytically and numerically. The linearly separable target rule is represented by an N-vector, whose time dependence is modelled by a random or deterministic drift process. A single-layer network is trained online using

Nonlinear adaptive inverse control via the unified model neural network

Science.gov (United States)

Jeng, Jin-Tsong; Lee, Tsu-Tian

1999-03-01

In this paper, we propose a new nonlinear adaptive inverse control via a unified model neural network. In order to overcome nonsystematic design and long training time in nonlinear adaptive inverse control, we propose the approximate transformable technique to obtain a Chebyshev Polynomials Based Unified Model (CPBUM) neural network for the feedforward/recurrent neural networks. It turns out that the proposed method can use less training time to get an inverse model. Finally, we apply this proposed method to control magnetic bearing system. The experimental results show that the proposed nonlinear adaptive inverse control architecture provides a greater flexibility and better performance in controlling magnetic bearing systems.

ACO-Initialized Wavelet Neural Network for Vibration Fault Diagnosis of Hydroturbine Generating Unit

Directory of Open Access Journals (Sweden)

Zhihuai Xiao

2015-01-01

Full Text Available Considering the drawbacks of traditional wavelet neural network, such as low convergence speed and high sensitivity to initial parameters, an ant colony optimization- (ACO- initialized wavelet neural network is proposed in this paper for vibration fault diagnosis of a hydroturbine generating unit. In this method, parameters of the wavelet neural network are initialized by the ACO algorithm, and then the wavelet neural network is trained by the gradient descent algorithm. Amplitudes of the frequency components of the hydroturbine generating unit vibration signals are used as feature vectors for wavelet neural network training to realize mapping relationship from vibration features to fault types. A real vibration fault diagnosis case result of a hydroturbine generating unit shows that the proposed method has faster convergence speed and stronger generalization ability than the traditional wavelet neural network and ACO wavelet neural network. Thus it can provide an effective solution for online vibration fault diagnosis of a hydroturbine generating unit.

A neural network model for credit risk evaluation.

Science.gov (United States)

Khashman, Adnan

2009-08-01

Credit scoring is one of the key analytical techniques in credit risk evaluation which has been an active research area in financial risk management. This paper presents a credit risk evaluation system that uses a neural network model based on the back propagation learning algorithm. We train and implement the neural network to decide whether to approve or reject a credit application, using seven learning schemes and real world credit applications from the Australian credit approval datasets. A comparison of the system performance under the different learning schemes is provided, furthermore, we compare the performance of two neural networks; with one and two hidden layers following the ideal learning scheme. Experimental results suggest that neural networks can be effectively used in automatic processing of credit applications.

A stochastic learning algorithm for layered neural networks

International Nuclear Information System (INIS)

Bartlett, E.B.; Uhrig, R.E.

1992-01-01

The random optimization method typically uses a Gaussian probability density function (PDF) to generate a random search vector. In this paper the random search technique is applied to the neural network training problem and is modified to dynamically seek out the optimal probability density function (OPDF) from which to select the search vector. The dynamic OPDF search process, combined with an auto-adaptive stratified sampling technique and a dynamic node architecture (DNA) learning scheme, completes the modifications of the basic method. The DNA technique determines the appropriate number of hidden nodes needed for a given training problem. By using DNA, researchers do not have to set the neural network architectures before training is initiated. The approach is applied to networks of generalized, fully interconnected, continuous perceptions. Computer simulation results are given

Use of neural networks to monitor power plant components

International Nuclear Information System (INIS)

Ikonomopoulos, A.; Tsoukalas, L.H.

1992-01-01

A new methodology is presented for nondestructive evaluation (NDE) of check valve performance and degradation. Artificial neural network (ANN) technology is utilized for processing frequency domain signatures of check valves operating in a nuclear power plant (NPP). Acoustic signatures obtained from different locations on a check valve are transformed from the time domain to the frequency domain and then used as input to a pretrained neural network. The neural network has been trained with data sets corresponding to normal operation, therefore establishing a basis for check valve satisfactory performance. Results obtained from the proposed methodology demonstrate the ability of neural networks to perform accurate and quick evaluations of check valve performance

HIV lipodystrophy case definition using artificial neural network modelling

DEFF Research Database (Denmark)

Ioannidis, John P A; Trikalinos, Thomas A; Law, Matthew

2003-01-01

OBJECTIVE: A case definition of HIV lipodystrophy has recently been developed from a combination of clinical, metabolic and imaging/body composition variables using logistic regression methods. We aimed to evaluate whether artificial neural networks could improve the diagnostic accuracy. METHODS......: The database of the case-control Lipodystrophy Case Definition Study was split into 504 subjects (265 with and 239 without lipodystrophy) used for training and 284 independent subjects (152 with and 132 without lipodystrophy) used for validation. Back-propagation neural networks with one or two middle layers...... were trained and validated. Results were compared against logistic regression models using the same information. RESULTS: Neural networks using clinical variables only (41 items) achieved consistently superior performance than logistic regression in terms of specificity, overall accuracy and area under...

TRIGA control rod position and reactivity transient Monitoring by Neural Networks

International Nuclear Information System (INIS)

Rosa, R.; Palomba, M.; Sepielli, M.

2008-01-01

Plant sensors drift or malfunction and operator actions in nuclear reactor control can be supported by sensor on-line monitoring, and data validation through soft-computing process. On-line recalibration can often avoid manual calibration or drifting component replacement. DSP requires prompt response to the modified conditions. Artificial Neural Network (ANN) and Fuzzy logic ensure: prompt response, link with field measurement and physical system behaviour, data incoming interpretation, and detection of discrepancy for mis-calibration or sensor faults. ANN (Artificial Neural Network) is a system based on the operation of biological neural networks. Although computing is day by day advancing, there are certain tasks that a program made for a common microprocessor is unable to perform. A software implementation of an ANN can be made with Pros and Cons. Pros: A neural network can perform tasks that a linear program can not; When an element of the neural network fails, it can continue without any problem by their parallel nature; A neural network learns and does not need to be reprogrammed; It can be implemented in any application; It can be implemented without any problem. Cons: The architecture of a neural network is different from the architecture of microprocessors therefore needs to be emulated; it requires high processing time for large neural networks; and the neural network needs training to operate. Three possibilities of training exist: Supervised learning: the network is trained providing input and matching output patterns; Unsupervised learning: input patterns are not a priori classified and the system must develop its own representation of the input stimuli; Reinforcement Learning: intermediate form of the above two types of learning, the learning machine does some action on the environment and gets a feedback response from the environment. Two TRIGAN ANN applications are considered: control rod position and fuel temperature. The outcome obtained in this

Neural network tagging in a toy model

International Nuclear Information System (INIS)

Milek, Marko; Patel, Popat

1999-01-01

The purpose of this study is a comparison of Artificial Neural Network approach to HEP analysis against the traditional methods. A toy model used in this analysis consists of two types of particles defined by four generic properties. A number of 'events' was created according to the model using standard Monte Carlo techniques. Several fully connected, feed forward multi layered Artificial Neural Networks were trained to tag the model events. The performance of each network was compared to the standard analysis mechanisms and significant improvement was observed

Artificial Neural Networks and Instructional Technology.

Science.gov (United States)

Carlson, Patricia A.

1991-01-01

Artificial neural networks (ANN), part of artificial intelligence, are discussed. Such networks are fed sample cases (training sets), learn how to recognize patterns in the sample data, and use this experience in handling new cases. Two cognitive roles for ANNs (intelligent filters and spreading, associative memories) are examined. Prototypes…

Training-Image Based Geostatistical Inversion Using a Spatial Generative Adversarial Neural Network

Science.gov (United States)

Laloy, Eric; Hérault, Romain; Jacques, Diederik; Linde, Niklas

2018-01-01

Probabilistic inversion within a multiple-point statistics framework is often computationally prohibitive for high-dimensional problems. To partly address this, we introduce and evaluate a new training-image based inversion approach for complex geologic media. Our approach relies on a deep neural network of the generative adversarial network (GAN) type. After training using a training image (TI), our proposed spatial GAN (SGAN) can quickly generate 2-D and 3-D unconditional realizations. A key characteristic of our SGAN is that it defines a (very) low-dimensional parameterization, thereby allowing for efficient probabilistic inversion using state-of-the-art Markov chain Monte Carlo (MCMC) methods. In addition, available direct conditioning data can be incorporated within the inversion. Several 2-D and 3-D categorical TIs are first used to analyze the performance of our SGAN for unconditional geostatistical simulation. Training our deep network can take several hours. After training, realizations containing a few millions of pixels/voxels can be produced in a matter of seconds. This makes it especially useful for simulating many thousands of realizations (e.g., for MCMC inversion) as the relative cost of the training per realization diminishes with the considered number of realizations. Synthetic inversion case studies involving 2-D steady state flow and 3-D transient hydraulic tomography with and without direct conditioning data are used to illustrate the effectiveness of our proposed SGAN-based inversion. For the 2-D case, the inversion rapidly explores the posterior model distribution. For the 3-D case, the inversion recovers model realizations that fit the data close to the target level and visually resemble the true model well.

Inverting radiometric measurements with a neural network

Science.gov (United States)

Measure, Edward M.; Yee, Young P.; Balding, Jeff M.; Watkins, Wendell R.

1992-02-01

A neural network scheme for retrieving remotely sensed vertical temperature profiles was applied to observed ground based radiometer measurements. The neural network used microwave radiance measurements and surface measurements of temperature and pressure as inputs. Because the microwave radiometer is capable of measuring 4 oxygen channels at 5 different elevation angles (9, 15, 25, 40, and 90 degs), 20 microwave measurements are potentially available. Because these measurements have considerable redundancy, a neural network was experimented with, accepting as inputs microwave measurements taken at 53.88 GHz, 40 deg; 57.45 GHz, 40 deg; and 57.45, 90 deg. The primary test site was located at White Sands Missile Range (WSMR), NM. Results are compared with measurements made simultaneously with balloon borne radiosonde instruments and with radiometric temperature retrievals made using more conventional retrieval algorithms. The neural network was trained using a Widrow-Hoff delta rule procedure. Functions of date to include season dependence in the retrieval process and functions of time to include diurnal effects were used as inputs to the neural network.

Application of neural networks to waste site screening

International Nuclear Information System (INIS)

Dabiri, A.E.; Garrett, M.; Kraft, T.; Hilton, J.; VanHammersveld, M.

1993-02-01

Waste site screening requires knowledge of the actual concentrations of hazardous materials and rates of flow around and below the site with time. The present approach consists primarily of drilling boreholes near contaminated sites and chemically analyzing the extracted physical samples and processing the data. This is expensive and time consuming. The feasibility of using neural network techniques to reduce the cost of waste site screening was investigated. Two neural network techniques, gradient descent back propagation and fully recurrent back propagation were utilized. The networks were trained with data received from Westinghouse Hanford Corporation. The results indicate that the network trained with the fully recurrent technique shows satisfactory generalization capability. The predicted results are close to the results obtained from a mathematical flow prediction model. It is possible to develop a new tool to predict the waste plume, thus substantially reducing the number of the bore sites and samplings. There are a variety of applications for this technique in environmental site screening and remediation. One of the obvious applications would be for optimum well siting. A neural network trained from the existing sampling data could be utilized to decide where would be the best position for the next bore site. Other applications are discussed in the report

Morphological neural networks

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.

Intrusion Detection System Using Deep Neural Network for In-Vehicle Network Security.

Directory of Open Access Journals (Sweden)

Min-Joo Kang

Full Text Available A novel intrusion detection system (IDS using a deep neural network (DNN is proposed to enhance the security of in-vehicular network. The parameters building the DNN structure are trained with probability-based feature vectors that are extracted from the in-vehicular network packets. For a given packet, the DNN provides the probability of each class discriminating normal and attack packets, and, thus the sensor can identify any malicious attack to the vehicle. As compared to the traditional artificial neural network applied to the IDS, the proposed technique adopts recent advances in deep learning studies such as initializing the parameters through the unsupervised pre-training of deep belief networks (DBN, therefore improving the detection accuracy. It is demonstrated with experimental results that the proposed technique can provide a real-time response to the attack with a significantly improved detection ratio in controller area network (CAN bus.

Intrusion Detection System Using Deep Neural Network for In-Vehicle Network Security.

Science.gov (United States)

Kang, Min-Joo; Kang, Je-Won

2016-01-01

A novel intrusion detection system (IDS) using a deep neural network (DNN) is proposed to enhance the security of in-vehicular network. The parameters building the DNN structure are trained with probability-based feature vectors that are extracted from the in-vehicular network packets. For a given packet, the DNN provides the probability of each class discriminating normal and attack packets, and, thus the sensor can identify any malicious attack to the vehicle. As compared to the traditional artificial neural network applied to the IDS, the proposed technique adopts recent advances in deep learning studies such as initializing the parameters through the unsupervised pre-training of deep belief networks (DBN), therefore improving the detection accuracy. It is demonstrated with experimental results that the proposed technique can provide a real-time response to the attack with a significantly improved detection ratio in controller area network (CAN) bus.

Cotton genotypes selection through artificial neural networks.

Science.gov (United States)

Júnior, E G Silva; Cardoso, D B O; Reis, M C; Nascimento, A F O; Bortolin, D I; Martins, M R; Sousa, L B

2017-09-27

Breeding programs currently use statistical analysis to assist in the identification of superior genotypes at various stages of a cultivar's development. Differently from these analyses, the computational intelligence approach has been little explored in genetic improvement of cotton. Thus, this study was carried out with the objective of presenting the use of artificial neural networks as auxiliary tools in the improvement of the cotton to improve fiber quality. To demonstrate the applicability of this approach, this research was carried out using the evaluation data of 40 genotypes. In order to classify the genotypes for fiber quality, the artificial neural networks were trained with replicate data of 20 genotypes of cotton evaluated in the harvests of 2013/14 and 2014/15, regarding fiber length, uniformity of length, fiber strength, micronaire index, elongation, short fiber index, maturity index, reflectance degree, and fiber quality index. This quality index was estimated by means of a weighted average on the determined score (1 to 5) of each characteristic of the HVI evaluated, according to its industry standards. The artificial neural networks presented a high capacity of correct classification of the 20 selected genotypes based on the fiber quality index, so that when using fiber length associated with the short fiber index, fiber maturation, and micronaire index, the artificial neural networks presented better results than using only fiber length and previous associations. It was also observed that to submit data of means of new genotypes to the neural networks trained with data of repetition, provides better results of classification of the genotypes. When observing the results obtained in the present study, it was verified that the artificial neural networks present great potential to be used in the different stages of a genetic improvement program of the cotton, aiming at the improvement of the fiber quality of the future cultivars.

Design and Implementation of Behavior Recognition System Based on Convolutional Neural Network

Directory of Open Access Journals (Sweden)

Yu Bo

2017-01-01

Full Text Available We build a set of human behavior recognition system based on the convolution neural network constructed for the specific human behavior in public places. Firstly, video of human behavior data set will be segmented into images, then we process the images by the method of background subtraction to extract moving foreground characters of body. Secondly, the training data sets are trained into the designed convolution neural network, and the depth learning network is constructed by stochastic gradient descent. Finally, the various behaviors of samples are classified and identified with the obtained network model, and the recognition results are compared with the current mainstream methods. The result show that the convolution neural network can study human behavior model automatically and identify human’s behaviors without any manually annotated trainings.

Disorder generated by interacting neural networks: application to econophysics and cryptography

International Nuclear Information System (INIS)

Kinzel, Wolfgang; Kanter, Ido

2003-01-01

When neural networks are trained on their own output signals they generate disordered time series. In particular, when two neural networks are trained on their mutual output they can synchronize; they relax to a time-dependent state with identical synaptic weights. Two applications of this phenomenon are discussed for (a) econophysics and (b) cryptography. (a) When agents competing in a closed market (minority game) are using neural networks to make their decisions, the total system relaxes to a state of good performance. (b) Two partners communicating over a public channel can find a common secret key

Neural network decoder for quantum error correcting codes

Science.gov (United States)

Krastanov, Stefan; Jiang, Liang

Artificial neural networks form a family of extremely powerful - albeit still poorly understood - tools used in anything from image and sound recognition through text generation to, in our case, decoding. We present a straightforward Recurrent Neural Network architecture capable of deducing the correcting procedure for a quantum error-correcting code from a set of repeated stabilizer measurements. We discuss the fault-tolerance of our scheme and the cost of training the neural network for a system of a realistic size. Such decoders are especially interesting when applied to codes, like the quantum LDPC codes, that lack known efficient decoding schemes.

Neural Networks through Shared Maps in Mobile Devices

Directory of Open Access Journals (Sweden)

William Raveane

2014-12-01

Full Text Available We introduce a hybrid system composed of a convolutional neural network and a discrete graphical model for image recognition. This system improves upon traditional sliding window techniques for analysis of an image larger than the training data by effectively processing the full input scene through the neural network in less time. The final result is then inferred from the neural network output through energy minimization to reach a more precize localization than what traditional maximum value class comparisons yield. These results are apt for applying this process in a mobile device for real time image recognition.

Hindcasting of storm waves using neural networks

Digital Repository Service at National Institute of Oceanography (India)

Rao, S.; Mandal, S.

Department NN neural network net i weighted sum of the inputs of neuron i o k network output at kth output node P total number of training pattern s i output of neuron i t k target output at kth output node 1. Introduction Severe storms occur in Bay of Bengal...), forecasting of runoff (Crespo and Mora, 1993), concrete strength (Kasperkiewicz et al., 1995). The uses of neural network in the coastal the wave conditions will change from year to year, thus a proper statistical and climatological treatment requires several...

Combined Ozone Retrieval From METOP Sensors Using META-Training Of Deep Neural Networks

Science.gov (United States)

Felder, Martin; Sehnke, Frank; Kaifel, Anton

2013-12-01

The newest installment of our well-proven Neural Net- work Ozone Retrieval System (NNORSY) combines the METOP sensors GOME-2 and IASI with cloud information from AVHRR. Through the use of advanced meta- learning techniques like automatic feature selection and automatic architecture search applied to a set of deep neural networks, having at least two or three hidden layers, we have been able to avoid many technical issues normally encountered during the construction of such a joint retrieval system. This has been made possible by harnessing the processing power of modern consumer graphics cards with high performance graphic processors (GPU), which decreases training times by about two orders of magnitude. The system was trained on data from 2009 and 2010, including target ozone profiles from ozone sondes, ACE- FTS and MLS-AURA. To make maximum use of tropospheric information in the spectra, the data were partitioned into several sets of different cloud fraction ranges with the GOME-2 FOV, on which specialized retrieval networks are being trained. For the final ozone retrieval processing the different specialized networks are combined. The resulting retrieval system is very stable and does not show any systematic dependence on solar zenith angle, scan angle or sensor degradation. We present several sensitivity studies with regard to cloud fraction and target sensor type, as well as the performance in several latitude bands and with respect to independent validation stations. A visual cross-comparison against high-resolution ozone profiles from the KNMI EUMETSAT Ozone SAF product has also been performed and shows some distinctive features which we will briefly discuss. Overall, we demonstrate that a complex retrieval system can now be constructed with a minimum of ma- chine learning knowledge, using automated algorithms for many design decisions previously requiring expert knowledge. Provided sufficient training data and computation power of GPUs is available, the

A TLD dose algorithm using artificial neural networks

International Nuclear Information System (INIS)

Moscovitch, M.; Rotunda, J.E.; Tawil, R.A.; Rathbone, B.A.

1995-01-01

An artificial neural network was designed and used to develop a dose algorithm for a multi-element thermoluminescence dosimeter (TLD). The neural network architecture is based on the concept of functional links network (FLN). Neural network is an information processing method inspired by the biological nervous system. A dose algorithm based on neural networks is fundamentally different as compared to conventional algorithms, as it has the capability to learn from its own experience. The neural network algorithm is shown the expected dose values (output) associated with given responses of a multi-element dosimeter (input) many times. The algorithm, being trained that way, eventually is capable to produce its own unique solution to similar (but not exactly the same) dose calculation problems. For personal dosimetry, the output consists of the desired dose components: deep dose, shallow dose and eye dose. The input consists of the TL data obtained from the readout of a multi-element dosimeter. The neural network approach was applied to the Harshaw Type 8825 TLD, and was shown to significantly improve the performance of this dosimeter, well within the U.S. accreditation requirements for personnel dosimeters

SYNAPTIC DEPRESSION IN DEEP NEURAL NETWORKS FOR SPEECH PROCESSING.

Science.gov (United States)

Zhang, Wenhao; Li, Hanyu; Yang, Minda; Mesgarani, Nima

2016-03-01

A characteristic property of biological neurons is their ability to dynamically change the synaptic efficacy in response to variable input conditions. This mechanism, known as synaptic depression, significantly contributes to the formation of normalized representation of speech features. Synaptic depression also contributes to the robust performance of biological systems. In this paper, we describe how synaptic depression can be modeled and incorporated into deep neural network architectures to improve their generalization ability. We observed that when synaptic depression is added to the hidden layers of a neural network, it reduces the effect of changing background activity in the node activations. In addition, we show that when synaptic depression is included in a deep neural network trained for phoneme classification, the performance of the network improves under noisy conditions not included in the training phase. Our results suggest that more complete neuron models may further reduce the gap between the biological performance and artificial computing, resulting in networks that better generalize to novel signal conditions.

Face recognition: a convolutional neural-network approach.

Science.gov (United States)

Lawrence, S; Giles, C L; Tsoi, A C; Back, A D

1997-01-01

We present a hybrid neural-network for human face recognition which compares favourably with other methods. The system combines local image sampling, a self-organizing map (SOM) neural network, and a convolutional neural network. The SOM provides a quantization of the image samples into a topological space where inputs that are nearby in the original space are also nearby in the output space, thereby providing dimensionality reduction and invariance to minor changes in the image sample, and the convolutional neural network provides partial invariance to translation, rotation, scale, and deformation. The convolutional network extracts successively larger features in a hierarchical set of layers. We present results using the Karhunen-Loeve transform in place of the SOM, and a multilayer perceptron (MLP) in place of the convolutional network for comparison. We use a database of 400 images of 40 individuals which contains quite a high degree of variability in expression, pose, and facial details. We analyze the computational complexity and discuss how new classes could be added to the trained recognizer.

Static human face recognition using artificial neural networks

International Nuclear Information System (INIS)

Qamar, R.; Shah, S.H.; Javed-ur-Rehman

2003-01-01

This paper presents a novel method of human face recognition using digital computers. A digital PC camera is used to take the BMP images of the human faces. An artificial neural network using Back Propagation Algorithm is developed as a recognition engine. The BMP images of the faces serve as the input patterns for this engine. A software 'Face Recognition' has been developed to recognize the human faces for which it is trained. Once the neural network is trained for patterns of the faces, the software is able to detect and recognize them with success rate of about 97%. (author)

Predicting carbonate permeabilities from wireline logs using a back-propagation neural network

International Nuclear Information System (INIS)

Wiener, J.M.; Moll, R.F.; Rogers, J.A.

1991-01-01

This paper explores the applicability of using Neural Networks to aid in the determination of carbonate permeability from wireline logs. Resistivity, interval transit time, neutron porosity, and bulk density logs form Texaco's Stockyard Creek Oil field were used as input to a specially designed neural network to predict core permeabilities in this carbonate reservoir. Also of interest was the comparison of the neural network's results to those of standard statistical techniques. The process of developing the neural network for this problem has shown that a good understanding of the data is required when creating the training set from which the network learns. This network was trained to learn core permeabilities from raw and transformed log data using a hyperbolic tangent transfer function and a sum of squares global error function. Also, it required two hidden layers to solve this particular problem

Application of Artificial Neural Networks for Efficient High-Resolution 2D DOA Estimation

Directory of Open Access Journals (Sweden)

M. Agatonović

2012-12-01

Full Text Available A novel method to provide high-resolution Two-Dimensional Direction of Arrival (2D DOA estimation employing Artificial Neural Networks (ANNs is presented in this paper. The observed space is divided into azimuth and elevation sectors. Multilayer Perceptron (MLP neural networks are employed to detect the presence of a source in a sector while Radial Basis Function (RBF neural networks are utilized for DOA estimation. It is shown that a number of appropriately trained neural networks can be successfully used for the high-resolution DOA estimation of narrowband sources in both azimuth and elevation. The training time of each smaller network is significantly re¬duced as different training sets are used for networks in detection and estimation stage. By avoiding the spectral search, the proposed method is suitable for real-time ap¬plications as it provides DOA estimates in a matter of seconds. At the same time, it demonstrates the accuracy comparable to that of the super-resolution 2D MUSIC algorithm.

Application of genetic neural network in steam generator fault diagnosing

International Nuclear Information System (INIS)

Lin Xiaogong; Jiang Xingwei; Liu Tao; Shi Xiaocheng

2005-01-01

In the paper, a new algorithm which neural network and genetic algorithm are mixed is adopted, aiming at the problems of slow convergence rate and easily falling into part minimums in network studying of traditional BP neural network, and used in the fault diagnosis of steam generator. The result shows that this algorithm can solve the convergence problem in the network trains effectively. (author)

Influence of the Training Methods in the Diagnosis of Multiple Sclerosis Using Radial Basis Functions Artificial Neural Networks

Directory of Open Access Journals (Sweden)

Ángel Gutiérrez

2015-04-01

Full Text Available The data available in the average clinical study of a disease is very often small. This is one of the main obstacles in the application of neural networks to the classification of biological signals used for diagnosing diseases. A rule of thumb states that the number of parameters (weights that can be used for training a neural network should be around 15% of the available data, to avoid overlearning. This condition puts a limit on the dimension of the input space. Different authors have used different approaches to solve this problem, like eliminating redundancy in the data, preprocessing the data to find centers for the radial basis functions, or extracting a small number of features that were used as inputs. It is clear that the classification would be better the more features we could feed into the network. The approach utilized in this paper is incrementing the number of training elements with randomly expanding training sets. This way the number of original signals does not constraint the dimension of the input set in the radial basis network. Then we train the network using the method that minimizes the error function using the gradient descent algorithm and the method that uses the particle swarm optimization technique. A comparison between the two methods showed that for the same number of iterations on both methods, the particle swarm optimization was faster, it was learning to recognize only the sick people. On the other hand, the gradient method was not as good in general better at identifying those people.

Pre-trained convolutional neural networks as feature extractors for tuberculosis detection.

Science.gov (United States)

Lopes, U K; Valiati, J F

2017-10-01

It is estimated that in 2015, approximately 1.8 million people infected by tuberculosis died, most of them in developing countries. Many of those deaths could have been prevented if the disease had been detected at an earlier stage, but the most advanced diagnosis methods are still cost prohibitive for mass adoption. One of the most popular tuberculosis diagnosis methods is the analysis of frontal thoracic radiographs; however, the impact of this method is diminished by the need for individual analysis of each radiography by properly trained radiologists. Significant research can be found on automating diagnosis by applying computational techniques to medical images, thereby eliminating the need for individual image analysis and greatly diminishing overall costs. In addition, recent improvements on deep learning accomplished excellent results classifying images on diverse domains, but its application for tuberculosis diagnosis remains limited. Thus, the focus of this work is to produce an investigation that will advance the research in the area, presenting three proposals to the application of pre-trained convolutional neural networks as feature extractors to detect the disease. The proposals presented in this work are implemented and compared to the current literature. The obtained results are competitive with published works demonstrating the potential of pre-trained convolutional networks as medical image feature extractors. Copyright © 2017 Elsevier Ltd. All rights reserved.

Practical Application of Neural Networks in State Space Control

DEFF Research Database (Denmark)

Bendtsen, Jan Dimon

the networks, although some modifications are needed for the method to apply to the multilayer perceptron network. In connection with the multilayer perceptron networks it is also pointed out how instantaneous, sample-by-sample linearized state space models can be extracted from a trained network, thus opening......In the present thesis we address some problems in discrete-time state space control of nonlinear dynamical systems and attempt to solve them using generic nonlinear models based on artificial neural networks. The main aim of the work is to examine how well such control algorithms perform when...... theoretic notions followed by a detailed description of the topology, neuron functions and learning rules of the two types of neural networks treated in the thesis, the multilayer perceptron and the neurofuzzy networks. In both cases, a Least Squares second-order gradient method is used to train...

Maximum entropy methods for extracting the learned features of deep neural networks.

Science.gov (United States)

Finnegan, Alex; Song, Jun S

2017-10-01

New architectures of multilayer artificial neural networks and new methods for training them are rapidly revolutionizing the application of machine learning in diverse fields, including business, social science, physical sciences, and biology. Interpreting deep neural networks, however, currently remains elusive, and a critical challenge lies in understanding which meaningful features a network is actually learning. We present a general method for interpreting deep neural networks and extracting network-learned features from input data. We describe our algorithm in the context of biological sequence analysis. Our approach, based on ideas from statistical physics, samples from the maximum entropy distribution over possible sequences, anchored at an input sequence and subject to constraints implied by the empirical function learned by a network. Using our framework, we demonstrate that local transcription factor binding motifs can be identified from a network trained on ChIP-seq data and that nucleosome positioning signals are indeed learned by a network trained on chemical cleavage nucleosome maps. Imposing a further constraint on the maximum entropy distribution also allows us to probe whether a network is learning global sequence features, such as the high GC content in nucleosome-rich regions. This work thus provides valuable mathematical tools for interpreting and extracting learned features from feed-forward neural networks.

Evaluation of the Performance of Feedforward and Recurrent Neural Networks in Active Cancellation of Sound Noise

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.

Optimization of blanking process using neural network simulation

International Nuclear Information System (INIS)

Hambli, R.

2005-01-01

The present work describes a methodology using the finite element method and neural network simulation in order to predict the optimum punch-die clearance during sheet metal blanking processes. A damage model is used in order to describe crack initiation and propagation into the sheet. The proposed approach combines predictive finite element and neural network modeling of the leading blanking parameters. Numerical results obtained by finite element computation including damage and fracture modeling were utilized to train the developed simulation environment based on back propagation neural network modeling. The comparative study between the numerical results and the experimental ones shows the good agreement. (author)

UAV Trajectory Modeling Using Neural Networks

Science.gov (United States)

Xue, Min

2017-01-01

Large amount of small Unmanned Aerial Vehicles (sUAVs) are projected to operate in the near future. Potential sUAV applications include, but not limited to, search and rescue, inspection and surveillance, aerial photography and video, precision agriculture, and parcel delivery. sUAVs are expected to operate in the uncontrolled Class G airspace, which is at or below 500 feet above ground level (AGL), where many static and dynamic constraints exist, such as ground properties and terrains, restricted areas, various winds, manned helicopters, and conflict avoidance among sUAVs. How to enable safe, efficient, and massive sUAV operations at the low altitude airspace remains a great challenge. NASA's Unmanned aircraft system Traffic Management (UTM) research initiative works on establishing infrastructure and developing policies, requirement, and rules to enable safe and efficient sUAVs' operations. To achieve this goal, it is important to gain insights of future UTM traffic operations through simulations, where the accurate trajectory model plays an extremely important role. On the other hand, like what happens in current aviation development, trajectory modeling should also serve as the foundation for any advanced concepts and tools in UTM. Accurate models of sUAV dynamics and control systems are very important considering the requirement of the meter level precision in UTM operations. The vehicle dynamics are relatively easy to derive and model, however, vehicle control systems remain unknown as they are usually kept by manufactures as a part of intellectual properties. That brings challenges to trajectory modeling for sUAVs. How to model the vehicle's trajectories with unknown control system? This work proposes to use a neural network to model a vehicle's trajectory. The neural network is first trained to learn the vehicle's responses at numerous conditions. Once being fully trained, given current vehicle states, winds, and desired future trajectory, the neural

Smooth function approximation using neural networks.

Science.gov (United States)

Ferrari, Silvia; Stengel, Robert F

2005-01-01

An algebraic approach for representing multidimensional nonlinear functions by feedforward neural networks is presented. In this paper, the approach is implemented for the approximation of smooth batch data containing the function's input, output, and possibly, gradient information. The training set is associated to the network adjustable parameters by nonlinear weight equations. The cascade structure of these equations reveals that they can be treated as sets of linear systems. Hence, the training process and the network approximation properties can be investigated via linear algebra. Four algorithms are developed to achieve exact or approximate matching of input-output and/or gradient-based training sets. Their application to the design of forward and feedback neurocontrollers shows that algebraic training is characterized by faster execution speeds and better generalization properties than contemporary optimization techniques.

Nuclear power plant fault-diagnosis using artificial neural networks

International Nuclear Information System (INIS)

Kim, Keehoon; Aljundi, T.L.; Bartlett, E.B.

1992-01-01

Artificial neural networks (ANNs) have been applied to various fields due to their fault and noise tolerance and generalization characteristics. As an application to nuclear engineering, we apply neural networks to the early recognition of nuclear power plant operational transients. If a transient or accident occurs, the network will advise the plant operators in a timely manner. More importantly, we investigate the ability of the network to provide a measure of the confidence level in its diagnosis. In this research an ANN is trained to diagnose the status of the San Onofre Nuclear Generation Station using data obtained from the plant's training simulator. Stacked generalization is then applied to predict the error in the ANN diagnosis. The data used consisted of 10 scenarios that include typical design basis accidents as well as less severe transients. The results show that the trained network is capable of diagnosing all 10 instabilities as well as providing a measure of the level of confidence in its diagnoses

System Identification, Prediction, Simulation and Control with Neural Networks

DEFF Research Database (Denmark)

Sørensen, O.

1997-01-01

a Gauss-Newton search direction is applied. 3) Amongst numerous model types, often met in control applications, only the Non-linear ARMAX (NARMAX) model, representing input/output description, is examined. A simulated example confirms that a neural network has the potential to perform excellent System......The intention of this paper is to make a systematic examination of the possibilities of applying neural networks in those technical areas, which are familiar to a control engineer. In other words, the potential of neural networks in control applications is given higher priority than a detailed...... study of the networks themselves. With this end in view the following restrictions have been made: 1) Amongst numerous neural network structures, only the Multi Layer Perceptron (a feed-forward network) is applied. 2) Amongst numerous training algorithms, only the Recursive Prediction Error Method using...

Forecasting Flare Activity Using Deep Convolutional Neural Networks

Science.gov (United States)

Hernandez, T.

2017-12-01

Current operational flare forecasting relies on human morphological analysis of active regions and the persistence of solar flare activity through time (i.e. that the Sun will continue to do what it is doing right now: flaring or remaining calm). In this talk we present the results of applying deep Convolutional Neural Networks (CNNs) to the problem of solar flare forecasting. CNNs operate by training a set of tunable spatial filters that, in combination with neural layer interconnectivity, allow CNNs to automatically identify significant spatial structures predictive for classification and regression problems. We will start by discussing the applicability and success rate of the approach, the advantages it has over non-automated forecasts, and how mining our trained neural network provides a fresh look into the mechanisms behind magnetic energy storage and release.

Input data preprocessing method for exchange rate forecasting via neural network

Directory of Open Access Journals (Sweden)

Antić Dragan S.

2014-01-01

Full Text Available The aim of this paper is to present a method for neural network input parameters selection and preprocessing. The purpose of this network is to forecast foreign exchange rates using artificial intelligence. Two data sets are formed for two different economic systems. Each system is represented by six categories with 70 economic parameters which are used in the analysis. Reduction of these parameters within each category was performed by using the principal component analysis method. Component interdependencies are established and relations between them are formed. Newly formed relations were used to create input vectors of a neural network. The multilayer feed forward neural network is formed and trained using batch training. Finally, simulation results are presented and it is concluded that input data preparation method is an effective way for preprocessing neural network data. [Projekat Ministarstva nauke Republike Srbije, br.TR 35005, br. III 43007 i br. III 44006

Inverse kinematics problem in robotics using neural networks

Science.gov (United States)

Choi, Benjamin B.; Lawrence, Charles

1992-01-01

In this paper, Multilayer Feedforward Networks are applied to the robot inverse kinematic problem. The networks are trained with endeffector position and joint angles. After training, performance is measured by having the network generate joint angles for arbitrary endeffector trajectories. A 3-degree-of-freedom (DOF) spatial manipulator is used for the study. It is found that neural networks provide a simple and effective way to both model the manipulator inverse kinematics and circumvent the problems associated with algorithmic solution methods.

Evaluation and scoring of radiotherapy treatment plans using an artificial neural network

International Nuclear Information System (INIS)

Willoughby, Twyla R.; Starkschall, George; Janjan, Nora A.; Rosen, Isaac I.

1996-01-01

Purpose: The objective of this work was to demonstrate the feasibility of using an artificial neural network to predict the clinical evaluation of radiotherapy treatment plans. Methods and Materials: Approximately 150 treatment plans were developed for 16 patients who received external-beam radiotherapy for soft-tissue sarcomas of the lower extremity. Plans were assigned a figure of merit by a radiation oncologist using a five-point rating scale. Plan scoring was performed by a single physician to ensure consistency in rating. Dose-volume information extracted from a training set of 511 treatment plans on 14 patients was correlated to the physician-generated figure of merit using an artificial neural network. The neural network was tested with a test set of 19 treatment plans on two patients whose plans were not used in the training of the neural net. Results: Physician scoring of treatment plans was consistent to within one point on the rating scale 88% of the time. The neural net reproduced the physician scores in the training set to within one point approximately 90% of the time. It reproduced the physician scores in the test set to within one point approximately 83% of the time. Conclusions: An artificial neural network can be trained to generate a score for a treatment plan that can be correlated to a clinically-based figure of merit. The accuracy of the neural net in scoring plans compares well with the reproducibility of the clinical scoring. The system of radiotherapy treatment plan evaluation using an artificial neural network demonstrates promise as a method for generating a clinically relevant figure of merit

IMPLEMENTATION OF ARTIFICIAL NEURAL NETWORK FOR FACE RECOGNITION USING GABOR FEATURE EXTRACTION

Directory of Open Access Journals (Sweden)

Muthukannan K

2013-11-01

Full Text Available Face detection and recognition is the first step for many applications in various fields such as identification and is used as a key to enter into the various electronic devices, video surveillance, and human computer interface and image database management. This paper focuses on feature extraction in an image using Gabor filter and the extracted image feature vector is then given as an input to the neural network. The neural network is trained with the input data. The Gabor wavelet concentrates on the important components of the face including eye, mouth, nose, cheeks. The main requirement of this technique is the threshold, which gives privileged sensitivity. The threshold values are the feature vectors taken from the faces. These feature vectors are given into the feed forward neural network to train the network. Using the feed forward neural network as a classifier, the recognized and unrecognized faces are classified. This classifier attains a higher face deduction rate. By training more input vectors the system proves to be effective. The effectiveness of the proposed method is demonstrated by the experimental results.

Computing single step operators of logic programming in radial basis function neural networks

Science.gov (United States)

Hamadneh, Nawaf; Sathasivam, Saratha; Choon, Ong Hong

2014-07-01

Logic programming is the process that leads from an original formulation of a computing problem to executable programs. A normal logic program consists of a finite set of clauses. A valuation I of logic programming is a mapping from ground atoms to false or true. The single step operator of any logic programming is defined as a function (Tp:I→I). Logic programming is well-suited to building the artificial intelligence systems. In this study, we established a new technique to compute the single step operators of logic programming in the radial basis function neural networks. To do that, we proposed a new technique to generate the training data sets of single step operators. The training data sets are used to build the neural networks. We used the recurrent radial basis function neural networks to get to the steady state (the fixed point of the operators). To improve the performance of the neural networks, we used the particle swarm optimization algorithm to train the networks.

Computing single step operators of logic programming in radial basis function neural networks

Energy Technology Data Exchange (ETDEWEB)

Hamadneh, Nawaf; Sathasivam, Saratha; Choon, Ong Hong [School of Mathematical Sciences, Universiti Sains Malaysia, 11800 USM, Penang (Malaysia)

2014-07-10

Logic programming is the process that leads from an original formulation of a computing problem to executable programs. A normal logic program consists of a finite set of clauses. A valuation I of logic programming is a mapping from ground atoms to false or true. The single step operator of any logic programming is defined as a function (T{sub p}:I→I). Logic programming is well-suited to building the artificial intelligence systems. In this study, we established a new technique to compute the single step operators of logic programming in the radial basis function neural networks. To do that, we proposed a new technique to generate the training data sets of single step operators. The training data sets are used to build the neural networks. We used the recurrent radial basis function neural networks to get to the steady state (the fixed point of the operators). To improve the performance of the neural networks, we used the particle swarm optimization algorithm to train the networks.

Computing single step operators of logic programming in radial basis function neural networks

International Nuclear Information System (INIS)

Hamadneh, Nawaf; Sathasivam, Saratha; Choon, Ong Hong

2014-01-01

Logic programming is the process that leads from an original formulation of a computing problem to executable programs. A normal logic program consists of a finite set of clauses. A valuation I of logic programming is a mapping from ground atoms to false or true. The single step operator of any logic programming is defined as a function (T p :I→I). Logic programming is well-suited to building the artificial intelligence systems. In this study, we established a new technique to compute the single step operators of logic programming in the radial basis function neural networks. To do that, we proposed a new technique to generate the training data sets of single step operators. The training data sets are used to build the neural networks. We used the recurrent radial basis function neural networks to get to the steady state (the fixed point of the operators). To improve the performance of the neural networks, we used the particle swarm optimization algorithm to train the networks

Synthesis of recurrent neural networks for dynamical system simulation.

Science.gov (United States)

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.

Adaptive model predictive process control using neural networks

Science.gov (United States)

Buescher, K.L.; Baum, C.C.; Jones, R.D.

1997-08-19

A control system for controlling the output of at least one plant process output parameter is implemented by adaptive model predictive control using a neural network. An improved method and apparatus provides for sampling plant output and control input at a first sampling rate to provide control inputs at the fast rate. The MPC system is, however, provided with a network state vector that is constructed at a second, slower rate so that the input control values used by the MPC system are averaged over a gapped time period. Another improvement is a provision for on-line training that may include difference training, curvature training, and basis center adjustment to maintain the weights and basis centers of the neural in an updated state that can follow changes in the plant operation apart from initial off-line training data. 46 figs.

Artificial Neural Networks to Detect Risk of Type 2 Diabetes | Baha ...

African Journals Online (AJOL)

A multilayer feedforward architecture with backpropagation algorithm was designed using Neural Network Toolbox of Matlab. The network was trained using batch mode backpropagation with gradient descent and momentum. Best performed network identified during the training was 2 hidden layers of 6 and 3 neurons, ...

Designing neural networks that process mean values of random variables

International Nuclear Information System (INIS)

Barber, Michael J.; Clark, John W.

2014-01-01

We develop a class of neural networks derived from probabilistic models posed in the form of Bayesian networks. Making biologically and technically plausible assumptions about the nature of the probabilistic models to be represented in the networks, we derive neural networks exhibiting standard dynamics that require no training to determine the synaptic weights, that perform accurate calculation of the mean values of the relevant random variables, that can pool multiple sources of evidence, and that deal appropriately with ambivalent, inconsistent, or contradictory evidence. - Highlights: • High-level neural computations are specified by Bayesian belief networks of random variables. • Probability densities of random variables are encoded in activities of populations of neurons. • Top-down algorithm generates specific neural network implementation of given computation. • Resulting “neural belief networks” process mean values of random variables. • Such networks pool multiple sources of evidence and deal properly with inconsistent evidence

Designing neural networks that process mean values of random variables

Energy Technology Data Exchange (ETDEWEB)

Barber, Michael J. [AIT Austrian Institute of Technology, Innovation Systems Department, 1220 Vienna (Austria); Clark, John W. [Department of Physics and McDonnell Center for the Space Sciences, Washington University, St. Louis, MO 63130 (United States); Centro de Ciências Matemáticas, Universidade de Madeira, 9000-390 Funchal (Portugal)

2014-06-13

We develop a class of neural networks derived from probabilistic models posed in the form of Bayesian networks. Making biologically and technically plausible assumptions about the nature of the probabilistic models to be represented in the networks, we derive neural networks exhibiting standard dynamics that require no training to determine the synaptic weights, that perform accurate calculation of the mean values of the relevant random variables, that can pool multiple sources of evidence, and that deal appropriately with ambivalent, inconsistent, or contradictory evidence. - Highlights: • High-level neural computations are specified by Bayesian belief networks of random variables. • Probability densities of random variables are encoded in activities of populations of neurons. • Top-down algorithm generates specific neural network implementation of given computation. • Resulting “neural belief networks” process mean values of random variables. • Such networks pool multiple sources of evidence and deal properly with inconsistent evidence.

Accelerator and feedback control simulation using neural networks

International Nuclear Information System (INIS)

Nguyen, D.; Lee, M.; Sass, R.; Shoaee, H.

1991-05-01

Unlike present constant model feedback system, neural networks can adapt as the dynamics of the process changes with time. Using a process model, the ''Accelerator'' network is first trained to simulate the dynamics of the beam for a given beam line. This ''Accelerator'' network is then used to train a second ''Controller'' network which performs the control function. In simulation, the networks are used to adjust corrector magnetics to control the launch angle and position of the beam to keep it on the desired trajectory when the incoming beam is perturbed. 4 refs., 3 figs

neural network based model o work based model of an industrial oil

African Journals Online (AJOL)

eobe

technique. g, Neural Network Model, Regression, Mean Square Error, PID controller. ... during the training processes. An additio ... used to carry out simulation studies of the mode .... A two-layer feed-forward neural network with Matlab.

Comparison of four Adaboost algorithm based artificial neural networks in wind speed predictions

International Nuclear Information System (INIS)

Liu, Hui; Tian, Hong-qi; Li, Yan-fei; Zhang, Lei

2015-01-01

Highlights: • Four hybrid algorithms are proposed for the wind speed decomposition. • Adaboost algorithm is adopted to provide a hybrid training framework. • MLP neural networks are built to do the forecasting computation. • Four important network training algorithms are included in the MLP networks. • All the proposed hybrid algorithms are suitable for the wind speed predictions. - Abstract: The technology of wind speed prediction is important to guarantee the safety of wind power utilization. In this paper, four different hybrid methods are proposed for the high-precision multi-step wind speed predictions based on the Adaboost (Adaptive Boosting) algorithm and the MLP (Multilayer Perceptron) neural networks. In the hybrid Adaboost–MLP forecasting architecture, four important algorithms are adopted for the training and modeling of the MLP neural networks, including GD-ALR-BP algorithm, GDM-ALR-BP algorithm, CG-BP-FR algorithm and BFGS algorithm. The aim of the study is to investigate the promoted forecasting percentages of the MLP neural networks by the Adaboost algorithm’ optimization under various training algorithms. The hybrid models in the performance comparison include Adaboost–GD-ALR-BP–MLP, Adaboost–GDM-ALR-BP–MLP, Adaboost–CG-BP-FR–MLP, Adaboost–BFGS–MLP, GD-ALR-BP–MLP, GDM-ALR-BP–MLP, CG-BP-FR–MLP and BFGS–MLP. Two experimental results show that: (1) the proposed hybrid Adaboost–MLP forecasting architecture is effective for the wind speed predictions; (2) the Adaboost algorithm has promoted the forecasting performance of the MLP neural networks considerably; (3) among the proposed Adaboost–MLP forecasting models, the Adaboost–CG-BP-FR–MLP model has the best performance; and (4) the improved percentages of the MLP neural networks by the Adaboost algorithm decrease step by step with the following sequence of training algorithms as: GD-ALR-BP, GDM-ALR-BP, CG-BP-FR and BFGS

Optimal Parameter for the Training of Multilayer Perceptron Neural Networks by Using Hierarchical Genetic Algorithm

International Nuclear Information System (INIS)

Orozco-Monteagudo, Maykel; Taboada-Crispi, Alberto; Gutierrez-Hernandez, Liliana

2008-01-01

This paper deals with the controversial topic of the selection of the parameters of a genetic algorithm, in this case hierarchical, used for training of multilayer perceptron neural networks for the binary classification. The parameters to select are the crossover and mutation probabilities of the control and parametric genes and the permanency percent. The results can be considered as a guide for using this kind of algorithm.

ACO-Initialized Wavelet Neural Network for Vibration Fault Diagnosis of Hydroturbine Generating Unit

OpenAIRE

Xiao, Zhihuai; He, Xinying; Fu, Xiangqian; Malik, O. P.

2015-01-01

Considering the drawbacks of traditional wavelet neural network, such as low convergence speed and high sensitivity to initial parameters, an ant colony optimization- (ACO-) initialized wavelet neural network is proposed in this paper for vibration fault diagnosis of a hydroturbine generating unit. In this method, parameters of the wavelet neural network are initialized by the ACO algorithm, and then the wavelet neural network is trained by the gradient descent algorithm. Amplitudes of the fr...

Self-consistent determination of the spike-train power spectrum in a neural network with sparse connectivity

Directory of Open Access Journals (Sweden)

Benjamin eDummer

2014-09-01

Full Text Available A major source of random variability in cortical networks is the quasi-random arrival of presynaptic action potentials from many other cells. In network studies as well as in the study of the response properties of single cells embedded in a network, synaptic background input is often approximated by Poissonian spike trains. However, the output statistics of the cells is in most cases far from being Poisson. This is inconsistent with the assumption of similar spike-train statistics for pre- and postsynaptic cells in a recurrent network. Here we tackle this problem for the popular class of integrate-and-fire neurons and study a self-consistent statistics of input and output spectra of neural spike trains. Instead of actually using a large network, we use an iterative scheme, in which we simulate a single neuron over several generations. In each of these generations, the neuron is stimulated with surrogate stochastic input that has a similar statistics as the output of the previous generation. For the surrogate input, we employ two distinct approximations: (i a superposition of renewal spike trains with the same interspike interval density as observed in the previous generation and (ii a Gaussian current with a power spectrum proportional to that observed in the previous generation. For input parameters that correspond to balanced input in the network, both the renewal and the Gaussian iteration procedure converge quickly and yield comparable results for the self-consistent spike-train power spectrum. We compare our results to large-scale simulations of a random sparsely connected network of leaky integrate-and-fire neurons (Brunel, J. Comp. Neurosci. 2000 and show that in the asynchronous regime close to a state of balanced synaptic input from the network, our iterative schemes provide excellent approximations to the autocorrelation of spike trains in the recurrent network.

Application of neural network technology to setpoint control of a simulated reactor experiment loop

International Nuclear Information System (INIS)

Cordes, G.A.; Bryan, S.R.; Powell, R.H.; Chick, D.R.

1991-01-01

This paper describes the design, implementation, and application of artificial neural networks to achieve temperature and flow rate control for a simulation of a typical experiment loop in the Advanced Test Reactor (ATR) located at the Idaho National Engineering Laboratory (INEL). The goal of the project was to research multivariate, nonlinear control using neural networks. A loop simulation code was adapted for the project and used to create a training set and test the neural network controller for comparison with the existing loop controllers. The results for the best neural network design are documented and compared with existing loop controller action. The neural network was shown to be as accurate at loop control as the classical controllers in the operating region represented by the training set. 5 refs., 8 figs., 3 tabs

Detecting and diagnosing SSME faults using an autoassociative neural network topology

Science.gov (United States)

Ali, M.; Dietz, W. E.; Kiech, E. L.

1989-01-01

An effort is underway at the University of Tennessee Space Institute to develop diagnostic expert system methodologies based on the analysis of patterns of behavior of physical mechanisms. In this approach, fault diagnosis is conceptualized as the mapping or association of patterns of sensor data to patterns representing fault conditions. Neural networks are being investigated as a means of storing and retrieving fault scenarios. Neural networks offer several powerful features in fault diagnosis, including (1) general pattern matching capabilities, (2) resistance to noisy input data, (3) the ability to be trained by example, and (4) the potential for implementation on parallel computer architectures. This paper presents (1) an autoassociative neural network topology, i.e. the network input and output is identical when properly trained, and hence learning is unsupervised; (2) the training regimen used; and (3) the response of the system to inputs representing both previously observed and unkown fault scenarios. The effects of noise on the integrity of the diagnosis are also evaluated.

Neural-Network Control Of Prosthetic And Robotic Hands

Science.gov (United States)

Buckley, Theresa M.

1991-01-01

Electronic neural networks proposed for use in controlling robotic and prosthetic hands and exoskeletal or glovelike electromechanical devices aiding intact but nonfunctional hands. Specific to patient, who activates grasping motion by voice command, by mechanical switch, or by myoelectric impulse. Patient retains higher-level control, while lower-level control provided by neural network analogous to that of miniature brain. During training, patient teaches miniature brain to perform specialized, anthropomorphic movements unique to himself or herself.

Gear Fault Diagnosis Based on BP Neural Network

Science.gov (United States)

Huang, Yongsheng; Huang, Ruoshi

2018-03-01

Gear transmission is more complex, widely used in machinery fields, which form of fault has some nonlinear characteristics. This paper uses BP neural network to train the gear of four typical failure modes, and achieves satisfactory results. Tested by using test data, test results have an agreement with the actual results. The results show that the BP neural network can effectively solve the complex state of gear fault in the gear fault diagnosis.

Quantitative phase microscopy using deep neural networks

Science.gov (United States)

Li, Shuai; Sinha, Ayan; Lee, Justin; Barbastathis, George

2018-02-01

Deep learning has been proven to achieve ground-breaking accuracy in various tasks. In this paper, we implemented a deep neural network (DNN) to achieve phase retrieval in a wide-field microscope. Our DNN utilized the residual neural network (ResNet) architecture and was trained using the data generated by a phase SLM. The results showed that our DNN was able to reconstruct the profile of the phase target qualitatively. In the meantime, large error still existed, which indicated that our approach still need to be improved.

UAV Trajectory Modeling Using Neural Networks

Science.gov (United States)

Xue, Min

2017-01-01

Massive small unmanned aerial vehicles are envisioned to operate in the near future. While there are lots of research problems need to be addressed before dense operations can happen, trajectory modeling remains as one of the keys to understand and develop policies, regulations, and requirements for safe and efficient unmanned aerial vehicle operations. The fidelity requirement of a small unmanned vehicle trajectory model is high because these vehicles are sensitive to winds due to their small size and low operational altitude. Both vehicle control systems and dynamic models are needed for trajectory modeling, which makes the modeling a great challenge, especially considering the fact that manufactures are not willing to share their control systems. This work proposed to use a neural network approach for modelling small unmanned vehicle's trajectory without knowing its control system and bypassing exhaustive efforts for aerodynamic parameter identification. As a proof of concept, instead of collecting data from flight tests, this work used the trajectory data generated by a mathematical vehicle model for training and testing the neural network. The results showed great promise because the trained neural network can predict 4D trajectories accurately, and prediction errors were less than 2:0 meters in both temporal and spatial dimensions.

Analysis of wave directional spreading using neural networks

Digital Repository Service at National Institute of Oceanography (India)

Deo, M.C.; Gondane, D.S.; SanilKumar, V.

describes how a representative spreading parameter could be arrived at from easily available wave parameters such as significant wave height and average zero-cross wave period, using the technique of neural networks. It is shown that training of the network...

Improved algorithms for circuit fault diagnosis based on wavelet packet and neural network

International Nuclear Information System (INIS)

Zhang, W-Q; Xu, C

2008-01-01

In this paper, two improved BP neural network algorithms of fault diagnosis for analog circuit are presented through using optimal wavelet packet transform(OWPT) or incomplete wavelet packet transform(IWPT) as preprocessor. The purpose of preprocessing is to reduce the nodes in input layer and hidden layer of BP neural network, so that the neural network gains faster training and convergence speed. At first, we apply OWPT or IWPT to the response signal of circuit under test(CUT), and then calculate the normalization energy of each frequency band. The normalization energy is used to train the BP neural network to diagnose faulty components in the analog circuit. These two algorithms need small network size, while have faster learning and convergence speed. Finally, simulation results illustrate the two algorithms are effective for fault diagnosis

Development of an accident diagnosis system using a dynamic neural network for nuclear power plants

International Nuclear Information System (INIS)

Lee, Seung Jun; Kim, Jong Hyun; Seong, Poong Hyun

2004-01-01

In this work, an accident diagnosis system using the dynamic neural network is developed. In order to help the plant operators to quickly identify the problem, perform diagnosis and initiate recovery actions ensuring the safety of the plant, many operator support system and accident diagnosis systems have been developed. Neural networks have been recognized as a good method to implement an accident diagnosis system. However, conventional accident diagnosis systems that used neural networks did not consider a time factor sufficiently. If the neural network could be trained according to time, it is possible to perform more efficient and detailed accidents analysis. Therefore, this work suggests a dynamic neural network which has different features from existing dynamic neural networks. And a simple accident diagnosis system is implemented in order to validate the dynamic neural network. After training of the prototype, several accident diagnoses were performed. The results show that the prototype can detect the accidents correctly with good performances

Neutron spectrometry and dosimetry by means of evolutive neural networks

International Nuclear Information System (INIS)

Ortiz R, J.M.; Martinez B, M.R.; Vega C, H.R.

2008-01-01

The artificial neural networks and the genetic algorithms are two relatively new areas of research, which have been subject to a growing interest during the last years. Both models are inspired by the nature, however, the neural networks are interested in the learning of a single individual, which is defined as fenotypic learning, while the evolutionary algorithms are interested in the adaptation of a population to a changing environment, that which is defined as genotypic learning. Recently, the use of the technology of neural networks has been applied with success in the area of the nuclear sciences, mainly in the areas of neutron spectrometry and dosimetry. The structure (network topology), as well as the learning parameters of a neural network, are factors that contribute in a significant way with the acting of the same one, however, it has been observed that the investigators in this area, carry out the selection of the network parameters through the essay and error technique, that which produces neural networks of poor performance and low generalization capacity. From the revised sources, it has been observed that the use of the evolutionary algorithms, seen as search techniques, it has allowed him to be possible to evolve and to optimize different properties of the neural networks, just as the initialization of the synaptic weights, the network architecture or the training algorithms without the human intervention. The objective of the present work is focused in analyzing the intersection of the neural networks and the evolutionary algorithms, analyzing like it is that the same ones can be used to help in the design processes and training of a neural network, this is, in the good selection of the structural parameters and of network learning, improving its generalization capacity, in such way that the same one is able to reconstruct in an efficient way neutron spectra and to calculate equivalent doses starting from the counting rates of a Bonner sphere

Photon spectrometry utilizing neural networks

International Nuclear Information System (INIS)

Silveira, R.; Benevides, C.; Lima, F.; Vilela, E.

2015-01-01

Having in mind the time spent on the uneventful work of characterization of the radiation beams used in a ionizing radiation metrology laboratory, the Metrology Service of the Centro Regional de Ciencias Nucleares do Nordeste - CRCN-NE verified the applicability of artificial intelligence (artificial neural networks) to perform the spectrometry in photon fields. For this, was developed a multilayer neural network, as an application for the classification of patterns in energy, associated with a thermoluminescent dosimetric system (TLD-700 and TLD-600). A set of dosimeters was initially exposed to various well known medium energies, between 40 keV and 1.2 MeV, coinciding with the beams determined by ISO 4037 standard, for the dose of 10 mSv in the quantity Hp(10), on a chest phantom (ISO slab phantom) with the purpose of generating a set of training data for the neural network. Subsequently, a new set of dosimeters irradiated in unknown energies was presented to the network with the purpose to test the method. The methodology used in this work was suitable for application in the classification of energy beams, having obtained 100% of the classification performed. (authors)

Artificial neural networks for processing fluorescence spectroscopy data in skin cancer diagnostics

International Nuclear Information System (INIS)

Lenhardt, L; Zeković, I; Dramićanin, T; Dramićanin, M D

2013-01-01

Over the years various optical spectroscopic techniques have been widely used as diagnostic tools in the discrimination of many types of malignant diseases. Recently, synchronous fluorescent spectroscopy (SFS) coupled with chemometrics has been applied in cancer diagnostics. The SFS method involves simultaneous scanning of both emission and excitation wavelengths while keeping the interval of wavelengths (constant-wavelength mode) or frequencies (constant-energy mode) between them constant. This method is fast, relatively inexpensive, sensitive and non-invasive. Total synchronous fluorescence spectra of normal skin, nevus and melanoma samples were used as input for training of artificial neural networks. Two different types of artificial neural networks were trained, the self-organizing map and the feed-forward neural network. Histopathology results of investigated skin samples were used as the gold standard for network output. Based on the obtained classification success rate of neural networks, we concluded that both networks provided high sensitivity with classification errors between 2 and 4%. (paper)

Artificial neural network based approach to transmission lines protection

International Nuclear Information System (INIS)

Joorabian, M.

1999-05-01

The aim of this paper is to present and accurate fault detection technique for high speed distance protection using artificial neural networks. The feed-forward multi-layer neural network with the use of supervised learning and the common training rule of error back-propagation is chosen for this study. Information available locally at the relay point is passed to a neural network in order for an assessment of the fault location to be made. However in practice there is a large amount of information available, and a feature extraction process is required to reduce the dimensionality of the pattern vectors, whilst retaining important information that distinguishes the fault point. The choice of features is critical to the performance of the neural networks learning and operation. A significant feature in this paper is that an artificial neural network has been designed and tested to enhance the precision of the adaptive capabilities for distance protection

Wind power prediction based on genetic neural network

Science.gov (United States)

Zhang, Suhan

2017-04-01

The scale of grid connected wind farms keeps increasing. To ensure the stability of power system operation, make a reasonable scheduling scheme and improve the competitiveness of wind farm in the electricity generation market, it's important to accurately forecast the short-term wind power. To reduce the influence of the nonlinear relationship between the disturbance factor and the wind power, the improved prediction model based on genetic algorithm and neural network method is established. To overcome the shortcomings of long training time of BP neural network and easy to fall into local minimum and improve the accuracy of the neural network, genetic algorithm is adopted to optimize the parameters and topology of neural network. The historical data is used as input to predict short-term wind power. The effectiveness and feasibility of the method is verified by the actual data of a certain wind farm as an example.

SuperSpike: Supervised Learning in Multilayer Spiking Neural Networks.

Science.gov (United States)

Zenke, Friedemann; Ganguli, Surya

2018-04-13

A vast majority of computation in the brain is performed by spiking neural networks. Despite the ubiquity of such spiking, we currently lack an understanding of how biological spiking neural circuits learn and compute in vivo, as well as how we can instantiate such capabilities in artificial spiking circuits in silico. Here we revisit the problem of supervised learning in temporally coding multilayer spiking neural networks. First, by using a surrogate gradient approach, we derive SuperSpike, a nonlinear voltage-based three-factor learning rule capable of training multilayer networks of deterministic integrate-and-fire neurons to perform nonlinear computations on spatiotemporal spike patterns. Second, inspired by recent results on feedback alignment, we compare the performance of our learning rule under different credit assignment strategies for propagating output errors to hidden units. Specifically, we test uniform, symmetric, and random feedback, finding that simpler tasks can be solved with any type of feedback, while more complex tasks require symmetric feedback. In summary, our results open the door to obtaining a better scientific understanding of learning and computation in spiking neural networks by advancing our ability to train them to solve nonlinear problems involving transformations between different spatiotemporal spike time patterns.

Adaptive Control of Nonlinear Discrete-Time Systems by Using OS-ELM Neural Networks

Directory of Open Access Journals (Sweden)

Xiao-Li Li

2014-01-01

Full Text Available As a kind of novel feedforward neural network with single hidden layer, ELM (extreme learning machine neural networks are studied for the identification and control of nonlinear dynamic systems. The property of simple structure and fast convergence of ELM can be shown clearly. In this paper, we are interested in adaptive control of nonlinear dynamic plants by using OS-ELM (online sequential extreme learning machine neural networks. Based on data scope division, the problem that training process of ELM neural network is sensitive to the initial training data is also solved. According to the output range of the controlled plant, the data corresponding to this range will be used to initialize ELM. Furthermore, due to the drawback of conventional adaptive control, when the OS-ELM neural network is used for adaptive control of the system with jumping parameters, the topological structure of the neural network can be adjusted dynamically by using multiple model switching strategy, and an MMAC (multiple model adaptive control will be used to improve the control performance. Simulation results are included to complement the theoretical results.

Performance of artificial neural networks and genetical evolved artificial neural networks unfolding techniques

International Nuclear Information System (INIS)

Ortiz R, J. M.; Martinez B, M. R.; Vega C, H. R.; Gallego D, E.; Lorente F, A.; Mendez V, R.; Los Arcos M, J. M.; Guerrero A, J. E.

2011-01-01

With the Bonner spheres spectrometer neutron spectrum is obtained through an unfolding procedure. Monte Carlo methods, Regularization, Parametrization, Least-squares, and Maximum Entropy are some of the techniques utilized for unfolding. In the last decade methods based on Artificial Intelligence Technology have been used. Approaches based on Genetic Algorithms and Artificial Neural Networks (Ann) have been developed in order to overcome the drawbacks of previous techniques. Nevertheless the advantages of Ann still it has some drawbacks mainly in the design process of the network, vg the optimum selection of the architectural and learning Ann parameters. In recent years the use of hybrid technologies, combining Ann and genetic algorithms, has been utilized to. In this work, several Ann topologies were trained and tested using Ann and Genetically Evolved Artificial Neural Networks in the aim to unfold neutron spectra using the count rates of a Bonner sphere spectrometer. Here, a comparative study of both procedures has been carried out. (Author)

The Evaluation on Data Mining Methods of Horizontal Bar Training Based on BP Neural Network

Directory of Open Access Journals (Sweden)

Zhang Yanhui

2015-01-01

Full Text Available With the rapid development of science and technology, data analysis has become an indispensable part of people’s work and life. Horizontal bar training has multiple categories. It is an emphasis for the re-search of related workers that categories of the training and match should be reduced. The application of data mining methods is discussed based on the problem of reducing categories of horizontal bar training. The BP neural network is applied to the cluster analysis and the principal component analysis, which are used to evaluate horizontal bar training. Two kinds of data mining methods are analyzed from two aspects, namely the operational convenience of data mining and the rationality of results. It turns out that the principal component analysis is more suitable for data processing of horizontal bar training.

Neural Network Classifiers for Local Wind Prediction.

Science.gov (United States)

Kretzschmar, Ralf; Eckert, Pierre; Cattani, Daniel; Eggimann, Fritz

2004-05-01

This paper evaluates the quality of neural network classifiers for wind speed and wind gust prediction with prediction lead times between +1 and +24 h. The predictions were realized based on local time series and model data. The selection of appropriate input features was initiated by time series analysis and completed by empirical comparison of neural network classifiers trained on several choices of input features. The selected input features involved day time, yearday, features from a single wind observation device at the site of interest, and features derived from model data. The quality of the resulting classifiers was benchmarked against persistence for two different sites in Switzerland. The neural network classifiers exhibited superior quality when compared with persistence judged on a specific performance measure, hit and false-alarm rates.

Inverse problems in eddy current testing using neural network

Science.gov (United States)

Yusa, N.; Cheng, W.; Miya, K.

2000-05-01

Reconstruction of crack in conductive material is one of the most important issues in the field of eddy current testing. Although many attempts to reconstruct cracks have been made, most of them deal with only artificial cracks machined with electro-discharge. However, in the case of natural cracks like stress corrosion cracking or inter-granular attack, there must be contact region and therefore their conductivity is not necessarily zero. In this study, an attempt to reconstruct natural cracks using neural network is presented. The neural network was trained through numerical simulated data obtained by the fast forward solver that calculated unflawed potential data a priori to save computational time. The solver is based on A-φ method discretized by using FEM-BEM A natural crack was modeled as an area whose conductivity was less than that of a specimen. The distribution of conductivity in that area was reconstructed as well. It took much time to train the network, but the speed of reconstruction was extremely fast after once it was trained. Well-trained network gave good reconstruction result.

Application of neural networks to signal prediction in nuclear power plant

International Nuclear Information System (INIS)

Wan Joo Kim; Soon Heung Chang; Byung Ho Lee

1993-01-01

This paper describes the feasibility study of an artificial neural network for signal prediction. The purpose of signal prediction is to estimate the value of undetected next time step signal. As the prediction method, based on the idea of auto regression, a few previous signals are inputs to the artificial neural network and the signal value of next time step is estimated with the outputs of the network. The artificial neural network can be applied to the nonlinear system and answers in short time. The training algorithm is a modified backpropagation model, which can effectively reduce the training time. The target signal of the simulation is the steam generator water level, which is one of the important parameters in nuclear power plants. The simulation result shows that the predicted value follows the real trend well

Fault detection and diagnosis for complex multivariable processes using neural networks

International Nuclear Information System (INIS)

Weerasinghe, M.

1998-06-01

Development of a reliable fault diagnosis method for large-scale industrial plants is laborious and often difficult to achieve due to the complexity of the targeted systems. The main objective of this thesis is to investigate the application of neural networks to the diagnosis of non-catastrophic faults in an industrial nuclear fuel processing plant. The proposed methods were initially developed by application to a simulated chemical process prior to further validation on real industrial data. The diagnosis of faults at a single operating point is first investigated. Statistical data conditioning methods of data scaling and principal component analysis are investigated to facilitate fault classification and reduce the complexity of neural networks. Successful fault diagnosis was achieved with significantly smaller networks than using all process variables as network inputs. Industrial processes often manufacture at various operating points, but demonstrated applications of neural networks for fault diagnosis usually only consider a single (primary) operating point. Developing a standard neural network scheme for fault diagnosis at all operating points would be usually impractical due to the unavailability of suitable training data for less frequently used (secondary) operating points. To overcome this problem, the application of a single neural network for the diagnosis of faults operating at different points is investigated. The data conditioning followed the same techniques as used for the fault diagnosis of a single operating point. The results showed that a single neural network could be successfully used to diagnose faults at operating points other than that it is trained for, and the data conditioning significantly improved the classification. Artificial neural networks have been shown to be an effective tool for process fault diagnosis. However, a main criticism is that details of the procedures taken to reach the fault diagnosis decisions are embedded in

Process identification through modular neural networks and rule extraction (extended abstract)

NARCIS (Netherlands)

van der Zwaag, B.J.; Slump, Cornelis H.; Spaanenburg, L.; Blockeel, Hendrik; Denecker, Marc

2002-01-01

Monolithic neural networks may be trained from measured data to establish knowledge about the process. Unfortunately, this knowledge is not guaranteed to be found and – if at all – hard to extract. Modular neural networks are better suited for this purpose. Domain-ordered by topology, rule

Optical Calibration Process Developed for Neural-Network-Based Optical Nondestructive Evaluation Method

Science.gov (United States)

Decker, Arthur J.

2004-01-01

A completely optical calibration process has been developed at Glenn for calibrating a neural-network-based nondestructive evaluation (NDE) method. The NDE method itself detects very small changes in the characteristic patterns or vibration mode shapes of vibrating structures as discussed in many references. The mode shapes or characteristic patterns are recorded using television or electronic holography and change when a structure experiences, for example, cracking, debonds, or variations in fastener properties. An artificial neural network can be trained to be very sensitive to changes in the mode shapes, but quantifying or calibrating that sensitivity in a consistent, meaningful, and deliverable manner has been challenging. The standard calibration approach has been difficult to implement, where the response to damage of the trained neural network is compared with the responses of vibration-measurement sensors. In particular, the vibration-measurement sensors are intrusive, insufficiently sensitive, and not numerous enough. In response to these difficulties, a completely optical alternative to the standard calibration approach was proposed and tested successfully. Specifically, the vibration mode to be monitored for structural damage was intentionally contaminated with known amounts of another mode, and the response of the trained neural network was measured as a function of the peak-to-peak amplitude of the contaminating mode. The neural network calibration technique essentially uses the vibration mode shapes of the undamaged structure as standards against which the changed mode shapes are compared. The published response of the network can be made nearly independent of the contaminating mode, if enough vibration modes are used to train the net. The sensitivity of the neural network can be adjusted for the environment in which the test is to be conducted. The response of a neural network trained with measured vibration patterns for use on a vibration isolation

Neural networks to predict exosphere temperature corrections

Science.gov (United States)

Choury, Anna; Bruinsma, Sean; Schaeffer, Philippe

2013-10-01

Precise orbit prediction requires a forecast of the atmospheric drag force with a high degree of accuracy. Artificial neural networks are universal approximators derived from artificial intelligence and are widely used for prediction. This paper presents a method of artificial neural networking for prediction of the thermosphere density by forecasting exospheric temperature, which will be used by the semiempirical thermosphere Drag Temperature Model (DTM) currently developed. Artificial neural network has shown to be an effective and robust forecasting model for temperature prediction. The proposed model can be used for any mission from which temperature can be deduced accurately, i.e., it does not require specific training. Although the primary goal of the study was to create a model for 1 day ahead forecast, the proposed architecture has been generalized to 2 and 3 days prediction as well. The impact of artificial neural network predictions has been quantified for the low-orbiting satellite Gravity Field and Steady-State Ocean Circulation Explorer in 2011, and an order of magnitude smaller orbit errors were found when compared with orbits propagated using the thermosphere model DTM2009.

Application of particle swarm optimization to identify gamma spectrum with neural network

International Nuclear Information System (INIS)

Shi Dongsheng; Di Yuming; Zhou Chunlin

2007-01-01

In applying neural network to identification of gamma spectra back propagation (BP) algorithm is usually trapped to a local optimum and has a low speed of convergence, whereas particle swarm optimization (PSO) is advantageous in terms of globe optimal searching. In this paper, we propose a new algorithm for neural network training, i.e. combined BP and PSO optimization, or PSO-BP algorithm. Practical example shows that the new algorithm can overcome shortcomings of BP algorithm and the neural network trained by it has a high ability of generalization with identification result of 100% correctness. It can be used effectively and reliably to identify gamma spectra. (authors)

Container-code recognition system based on computer vision and deep neural networks

Science.gov (United States)

Liu, Yi; Li, Tianjian; Jiang, Li; Liang, Xiaoyao

2018-04-01

Automatic container-code recognition system becomes a crucial requirement for ship transportation industry in recent years. In this paper, an automatic container-code recognition system based on computer vision and deep neural networks is proposed. The system consists of two modules, detection module and recognition module. The detection module applies both algorithms based on computer vision and neural networks, and generates a better detection result through combination to avoid the drawbacks of the two methods. The combined detection results are also collected for online training of the neural networks. The recognition module exploits both character segmentation and end-to-end recognition, and outputs the recognition result which passes the verification. When the recognition module generates false recognition, the result will be corrected and collected for online training of the end-to-end recognition sub-module. By combining several algorithms, the system is able to deal with more situations, and the online training mechanism can improve the performance of the neural networks at runtime. The proposed system is able to achieve 93% of overall recognition accuracy.

Alpha spectral analysis via artificial neural networks

International Nuclear Information System (INIS)

Kangas, L.J.; Hashem, S.; Keller, P.E.; Kouzes, R.T.; Troyer, G.L.

1994-10-01

An artificial neural network system that assigns quality factors to alpha particle energy spectra is discussed. The alpha energy spectra are used to detect plutonium contamination in the work environment. The quality factors represent the levels of spectral degradation caused by miscalibration and foreign matter affecting the instruments. A set of spectra was labeled with a quality factor by an expert and used in training the artificial neural network expert system. The investigation shows that the expert knowledge of alpha spectra quality factors can be transferred to an ANN system

Neural network approach to radiologic lesion detection

International Nuclear Information System (INIS)

Newman, F.D.; Raff, U.; Stroud, D.

1989-01-01

An area of artificial intelligence that has gained recent attention is the neural network approach to pattern recognition. The authors explore the use of neural networks in radiologic lesion detection with what is known in the literature as the novelty filter. This filter uses a linear model; images of normal patterns become training vectors and are stored as columns of a matrix. An image of an abnormal pattern is introduced and the abnormality or novelty is extracted. A VAX 750 was used to encode the novelty filter, and two experiments have been examined

High-Dimensional Function Approximation With Neural Networks for Large Volumes of Data.

Science.gov (United States)

Andras, Peter

2018-02-01

Approximation of high-dimensional functions is a challenge for neural networks due to the curse of dimensionality. Often the data for which the approximated function is defined resides on a low-dimensional manifold and in principle the approximation of the function over this manifold should improve the approximation performance. It has been show that projecting the data manifold into a lower dimensional space, followed by the neural network approximation of the function over this space, provides a more precise approximation of the function than the approximation of the function with neural networks in the original data space. However, if the data volume is very large, the projection into the low-dimensional space has to be based on a limited sample of the data. Here, we investigate the nature of the approximation error of neural networks trained over the projection space. We show that such neural networks should have better approximation performance than neural networks trained on high-dimensional data even if the projection is based on a relatively sparse sample of the data manifold. We also find that it is preferable to use a uniformly distributed sparse sample of the data for the purpose of the generation of the low-dimensional projection. We illustrate these results considering the practical neural network approximation of a set of functions defined on high-dimensional data including real world data as well.

Tracking and vertex finding with drift chambers and neural networks

International Nuclear Information System (INIS)

Lindsey, C.

1991-09-01

Finding tracks, track vertices and event vertices with neural networks from drift chamber signals is discussed. Simulated feed-forward neural networks have been trained with back-propagation to give track parameters using Monte Carlo simulated tracks in one case and actual experimental data in another. Effects on network performance of limited weight resolution, noise and drift chamber resolution are given. Possible implementations in hardware are discussed. 7 refs., 10 figs

Development of Artificial Neural Network Model for Diesel Fuel Properties Prediction using Vibrational Spectroscopy.

Science.gov (United States)

Bolanča, Tomislav; Marinović, Slavica; Ukić, Sime; Jukić, Ante; Rukavina, Vinko

2012-06-01

This paper describes development of artificial neural network models which can be used to correlate and predict diesel fuel properties from several FTIR-ATR absorbances and Raman intensities as input variables. Multilayer feed forward and radial basis function neural networks have been used to rapid and simultaneous prediction of cetane number, cetane index, density, viscosity, distillation temperatures at 10% (T10), 50% (T50) and 90% (T90) recovery, contents of total aromatics and polycyclic aromatic hydrocarbons of commercial diesel fuels. In this study two-phase training procedures for multilayer feed forward networks were applied. While first phase training algorithm was constantly the back propagation one, two second phase training algorithms were varied and compared, namely: conjugate gradient and quasi Newton. In case of radial basis function network, radial layer was trained using K-means radial assignment algorithm and three different radial spread algorithms: explicit, isotropic and K-nearest neighbour. The number of hidden layer neurons and experimental data points used for the training set have been optimized for both neural networks in order to insure good predictive ability by reducing unnecessary experimental work. This work shows that developed artificial neural network models can determine main properties of diesel fuels simultaneously based on a single and fast IR or Raman measurement.

Identification of generalized state transfer matrix using neural networks

International Nuclear Information System (INIS)

Zhu Changchun

2001-01-01

The research is introduced on identification of generalized state transfer matrix of linear time-invariant (LTI) system by use of neural networks based on LM (Levenberg-Marquart) algorithm. Firstly, the generalized state transfer matrix is defined. The relationship between the identification of state transfer matrix of structural dynamics and the identification of the weight matrix of neural networks has been established in theory. A singular layer neural network is adopted to obtain the structural parameters as a powerful tool that has parallel distributed processing ability and the property of adaptation or learning. The constraint condition of weight matrix of the neural network is deduced so that the learning and training of the designed network can be more effective. The identified neural network can be used to simulate the structural response excited by any other signals. In order to cope with its further application in practical problems, some noise (5% and 10%) is expected to be present in the response measurements. Results from computer simulation studies show that this method is valid and feasible

Artificial neural networks in neutron dosimetry

International Nuclear Information System (INIS)

Vega-Carrillo, H. R.; Hernandez-Davila, V. M.; Manzanares-Acuna, E.; Mercado, G. A.; Gallego, E.; Lorente, A.; Perales-Munoz, W. A.; Robles-Rodriguez, J. A.

2006-01-01

An artificial neural network (ANN) has been designed to obtain neutron doses using only the count rates of a Bonner spheres spectrometer (BSS). Ambient, personal and effective neutron doses were included. One hundred and eighty-one neutron spectra were utilised to calculate the Bonner count rates and the neutron doses. The spectra were transformed from lethargy to energy distribution and were re-binned to 31 energy groups using the MCNP 4C code. Re-binned spectra, UTA4 response matrix and fluence-to-dose coefficients were used to calculate the count rates in the BSS and the doses. Count rates were used as input and the respective doses were used as output during neural network training. Training and testing were carried out in the MATLAB R environment. The impact of uncertainties in BSS count rates upon the dose quantities calculated with the ANN was investigated by modifying by ±5% the BSS count rates used in the training set. The use of ANNs in neutron dosimetry is an alternative procedure that overcomes the drawbacks associated with this ill-conditioned problem. (authors)

Forecasting solar proton event with artificial neural network

Science.gov (United States)

Gong, J.; Wang, J.; Xue, B.; Liu, S.; Zou, Z.

Solar proton event (SPE), relatively rare but popular in solar maximum, can bring hazard situation to spacecraft. As a special event, SPE always accompanies flare, which is also called proton flare. To produce such an eruptive event, large amount energy must be accumulated within the active region. So we can investigate the character of the active region and its evolving trend, together with other such as cm radio emission and soft X-ray background to evaluate the potential of SEP in chosen area. In order to summarize the omen of SPEs in the active regions behind the observed parameters, we employed AI technology. Full connecting neural network was chosen to fulfil this job. After constructing the network, we train it with 13 parameters that was able to exhibit the character of active regions and their evolution trend. More than 80 sets of event parameter were defined to teach the neural network to identify whether an active region was potential of SPE. Then we test this model with a data base consisting SPE and non-SPE cases that was not used to train the neural network. The result showed that 75% of the choice by the model was right.

Selected aspects of modelling of foreign exchange rates with neural networks

Directory of Open Access Journals (Sweden)

Václav Mastný

2005-01-01

Full Text Available This paper deals with forecasting of the high-frequency foreign exchange market with neural networks. The objective is to investigate some aspects of modelling with neural networks (impact of topology, size of training set and time horizon of the forecast on the performance of the network. The data used for the purpose of this paper contain 15-minute time series of US dollar against other major currencies, Japanese Yen, British Pound and Euro. The results show, that performance of the network in terms of correct directorial change is negatively influenced by increasing number of hidden neurons and decreasing size of training set. The performance of the network is influenced by sampling frequency.

Artificial Neural Network L* from different magnetospheric field models

Science.gov (United States)

Yu, Y.; Koller, J.; Zaharia, S. G.; Jordanova, V. K.

2011-12-01

The third adiabatic invariant L* plays an important role in modeling and understanding the radiation belt dynamics. The popular way to numerically obtain the L* value follows the recipe described by Roederer [1970], which is, however, slow and computational expensive. This work focuses on a new technique, which can compute the L* value in microseconds without losing much accuracy: artificial neural networks. Since L* is related to the magnetic flux enclosed by a particle drift shell, global magnetic field information needed to trace the drift shell is required. A series of currently popular empirical magnetic field models are applied to create the L* data pool using 1 million data samples which are randomly selected within a solar cycle and within the global magnetosphere. The networks, trained from the above L* data pool, can thereby be used for fairly efficient L* calculation given input parameters valid within the trained temporal and spatial range. Besides the empirical magnetospheric models, a physics-based self-consistent inner magnetosphere model (RAM-SCB) developed at LANL is also utilized to calculate L* values and then to train the L* neural network. This model better predicts the magnetospheric configuration and therefore can significantly improve the L*. The above neural network L* technique will enable, for the first time, comprehensive solar-cycle long studies of radiation belt processes. However, neural networks trained from different magnetic field models can result in different L* values, which could cause mis-interpretation of radiation belt dynamics, such as where the source of the radiation belt charged particle is and which mechanism is dominant in accelerating the particles. Such a fact calls for attention to cautiously choose a magnetospheric field model for the L* calculation.

Research of convolutional neural networks for traffic sign recognition

OpenAIRE

Stadalnikas, Kasparas

2017-01-01

In this thesis the convolutional neural networks application for traffic sign recognition is analyzed. Thesis describes the basic operations, techniques that are commonly used to apply in the image classification using convolutional neural networks. Also, this paper describes the data sets used for traffic sign recognition, their problems affecting the final training results. The paper reviews most popular existing technologies – frameworks for developing the solution for traffic sign recogni...

Eddy Current Flaw Characterization Using Neural Networks

International Nuclear Information System (INIS)

Song, S. J.; Park, H. J.; Shin, Y. K.

1998-01-01

Determination of location, shape and size of a flaw from its eddy current testing signal is one of the fundamental issues in eddy current nondestructive evaluation of steam generator tubes. Here, we propose an approach to this problem; an inversion of eddy current flaw signal using neural networks trained by finite element model-based synthetic signatures. Total 216 eddy current signals from four different types of axisymmetric flaws in tubes are generated by finite element models of which the accuracy is experimentally validated. From each simulated signature, total 24 eddy current features are extracted and among them 13 features are finally selected for flaw characterization. Based on these features, probabilistic neural networks discriminate flaws into four different types according to the location and the shape, and successively back propagation neural networks determine the size parameters of the discriminated flaw

Liquefaction Microzonation of Babol City Using Artificial Neural Network

DEFF Research Database (Denmark)

Farrokhzad, F.; Choobbasti, A.J.; Barari, Amin

2012-01-01

that will be less susceptible to damage during earthquakes. The scope of present study is to prepare the liquefaction microzonation map for the Babol city based on Seed and Idriss (1983) method using artificial neural network. Artificial neural network (ANN) is one of the artificial intelligence (AI) approaches...... microzonation map is produced for research area. Based on the obtained results, it can be stated that the trained neural network is capable in prediction of liquefaction potential with an acceptable level of confidence. At the end, zoning of the city is carried out based on the prediction of liquefaction...... that can be classified as machine learning. Simplified methods have been practiced by researchers to assess nonlinear liquefaction potential of soil. In order to address the collective knowledge built-up in conventional liquefaction engineering, an alternative general regression neural network model...

Classification of urine sediment based on convolution neural network

Science.gov (United States)

Pan, Jingjing; Jiang, Cunbo; Zhu, Tiantian

2018-04-01

By designing a new convolution neural network framework, this paper breaks the constraints of the original convolution neural network framework requiring large training samples and samples of the same size. Move and cropping the input images, generate the same size of the sub-graph. And then, the generated sub-graph uses the method of dropout, increasing the diversity of samples and preventing the fitting generation. Randomly select some proper subset in the sub-graphic set and ensure that the number of elements in the proper subset is same and the proper subset is not the same. The proper subsets are used as input layers for the convolution neural network. Through the convolution layer, the pooling, the full connection layer and output layer, we can obtained the classification loss rate of test set and training set. In the red blood cells, white blood cells, calcium oxalate crystallization classification experiment, the classification accuracy rate of 97% or more.

Predicting Student Academic Performance: A Comparison of Two Meta-Heuristic Algorithms Inspired by Cuckoo Birds for Training Neural Networks

Directory of Open Access Journals (Sweden)

Jeng-Fung Chen

2014-10-01

Full Text Available Predicting student academic performance with a high accuracy facilitates admission decisions and enhances educational services at educational institutions. This raises the need to propose a model that predicts student performance, based on the results of standardized exams, including university entrance exams, high school graduation exams, and other influential factors. In this study, an approach to the problem based on the artificial neural network (ANN with the two meta-heuristic algorithms inspired by cuckoo birds and their lifestyle, namely, Cuckoo Search (CS and Cuckoo Optimization Algorithm (COA is proposed. In particular, we used previous exam results and other factors, such as the location of the student’s high school and the student’s gender as input variables, and predicted the student academic performance. The standard CS and standard COA were separately utilized to train the feed-forward network for prediction. The algorithms optimized the weights between layers and biases of the neuron network. The simulation results were then discussed and analyzed to investigate the prediction ability of the neural network trained by these two algorithms. The findings demonstrated that both CS and COA have potential in training ANN and ANN-COA obtained slightly better results for predicting student academic performance in this case. It is expected that this work may be used to support student admission procedures and strengthen the service system in educational institutions.

Design of Jetty Piles Using Artificial Neural Networks

Directory of Open Access Journals (Sweden)

Yongjei Lee

2014-01-01

Full Text Available To overcome the complication of jetty pile design process, artificial neural networks (ANN are adopted. To generate the training samples for training ANN, finite element (FE analysis was performed 50 times for 50 different design cases. The trained ANN was verified with another FE analysis case and then used as a structural analyzer. The multilayer neural network (MBPNN with two hidden layers was used for ANN. The framework of MBPNN was defined as the input with the lateral forces on the jetty structure and the type of piles and the output with the stress ratio of the piles. The results from the MBPNN agree well with those from FE analysis. Particularly for more complex modes with hundreds of different design cases, the MBPNN would possibly substitute parametric studies with FE analysis saving design time and cost.

Neural network based approach for tuning of SNS feedback and feedforward controllers

International Nuclear Information System (INIS)

Kwon, Sung-Il; Prokop, Mark S.; Regan, Amy H.

2002-01-01

The primary controllers in the SNS low level RF system are proportional-integral (PI) feedback controllers. To obtain the best performance of the linac control systems, approximately 91 individual PI controller gains should be optimally tuned. Tuning is time consuming and requires automation. In this paper, a neural network is used for the controller gain tuning. A neural network can approximate any continuous mapping through learning. In a sense, the cavity loop PI controller is a continuous mapping of the tracking error and its one-sample-delay inputs to the controller output. Also, monotonic cavity output with respect to its input makes knowing the detailed parameters of the cavity unnecessary. Hence the PI controller is a prime candidate for approximation through a neural network. Using mean square error minimization to train the neural network along with a continuous mapping of appropriate weights, optimally tuned PI controller gains can be determined. The same neural network approximation property is also applied to enhance the adaptive feedforward controller performance. This is done by adjusting the feedforward controller gains, forgetting factor, and learning ratio. Lastly, the automation of the tuning procedure data measurement, neural network training, tuning and loading the controller gain to the DSP is addressed.

Tomographic image reconstruction using Artificial Neural Networks

International Nuclear Information System (INIS)

Paschalis, P.; Giokaris, N.D.; Karabarbounis, A.; Loudos, G.K.; Maintas, D.; Papanicolas, C.N.; Spanoudaki, V.; Tsoumpas, Ch.; Stiliaris, E.

2004-01-01

A new image reconstruction technique based on the usage of an Artificial Neural Network (ANN) is presented. The most crucial factor in designing such a reconstruction system is the network architecture and the number of the input projections needed to reconstruct the image. Although the training phase requires a large amount of input samples and a considerable CPU time, the trained network is characterized by simplicity and quick response. The performance of this ANN is tested using several image patterns. It is intended to be used together with a phantom rotating table and the γ-camera of IASA for SPECT image reconstruction

Neural network versus classical time series forecasting models

Science.gov (United States)

Nor, Maria Elena; Safuan, Hamizah Mohd; Shab, Noorzehan Fazahiyah Md; Asrul, Mohd; Abdullah, Affendi; Mohamad, Nurul Asmaa Izzati; Lee, Muhammad Hisyam

2017-05-01

Artificial neural network (ANN) has advantage in time series forecasting as it has potential to solve complex forecasting problems. This is because ANN is data driven approach which able to be trained to map past values of a time series. In this study the forecast performance between neural network and classical time series forecasting method namely seasonal autoregressive integrated moving average models was being compared by utilizing gold price data. Moreover, the effect of different data preprocessing on the forecast performance of neural network being examined. The forecast accuracy was evaluated using mean absolute deviation, root mean square error and mean absolute percentage error. It was found that ANN produced the most accurate forecast when Box-Cox transformation was used as data preprocessing.

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)

Identification of illicit drugs by using SOM neural networks

Energy Technology Data Exchange (ETDEWEB)

Liang Meiyan; Shen Jingling; Wang Guangqin [Beijing Key Lab for Terahertz Spectroscopy and Imaging, Key Laboratory of Terahertz Optoelectronics, Ministry of Education, Department of Physics, Capital Normal University, Beijing 100037 (China)], E-mail: liangyan661982@163.com, E-mail: jinglingshen@gmail.com, E-mail: pywgq2004@163.com

2008-07-07

Absorption spectra of six illicit drugs were measured by using the terahertz time-domain spectroscopy technique in the range 0.2-2.6 THz and then clustered with self-organization feature map (SOM) artificial neural network. After the network training process, the spectra collected at another time were identified successfully by the well-trained SOM network. An effective distance was introduced as a quantitative criterion to decide which cluster the new spectra were affiliated with.

Identification of illicit drugs by using SOM neural networks

International Nuclear Information System (INIS)

Liang Meiyan; Shen Jingling; Wang Guangqin

2008-01-01

Absorption spectra of six illicit drugs were measured by using the terahertz time-domain spectroscopy technique in the range 0.2-2.6 THz and then clustered with self-organization feature map (SOM) artificial neural network. After the network training process, the spectra collected at another time were identified successfully by the well-trained SOM network. An effective distance was introduced as a quantitative criterion to decide which cluster the new spectra were affiliated with

The application of artificial neural network in radon disaster model of uranium mining

International Nuclear Information System (INIS)

Zhu Yufeng; Zhu Guogen; Zhou Shijian

2012-01-01

The structural features, data analysis and learning process of feed-forward neural network (BP ANN) were analyzed at first. Rodon sample from Fuzhou Jinan Uranium Industry Limited Company were used to training the network and make the forecast then, and a forecasting model was established for the radon disaster in uranium mines. The method and effectiveness of BP neural network in predicting radon disaster was discussed. The test of training samples showed that the BP network had gotten fairly satisfied result in predicting mine radon disaster. (authors)

Plant species classification using deep convolutional neural network

DEFF Research Database (Denmark)

Dyrmann, Mads; Karstoft, Henrik; Midtiby, Henrik Skov

2016-01-01

Information on which weed species are present within agricultural fields is important for site specific weed management. This paper presents a method that is capable of recognising plant species in colour images by using a convolutional neural network. The network is built from scratch trained an...

Short-Term Load Forecasting Model Based on Quantum Elman Neural Networks

Directory of Open Access Journals (Sweden)

Zhisheng Zhang

2016-01-01

Full Text Available Short-term load forecasting model based on quantum Elman neural networks was constructed in this paper. The quantum computation and Elman feedback mechanism were integrated into quantum Elman neural networks. Quantum computation can effectively improve the approximation capability and the information processing ability of the neural networks. Quantum Elman neural networks have not only the feedforward connection but also the feedback connection. The feedback connection between the hidden nodes and the context nodes belongs to the state feedback in the internal system, which has formed specific dynamic memory performance. Phase space reconstruction theory is the theoretical basis of constructing the forecasting model. The training samples are formed by means of K-nearest neighbor approach. Through the example simulation, the testing results show that the model based on quantum Elman neural networks is better than the model based on the quantum feedforward neural network, the model based on the conventional Elman neural network, and the model based on the conventional feedforward neural network. So the proposed model can effectively improve the prediction accuracy. The research in the paper makes a theoretical foundation for the practical engineering application of the short-term load forecasting model based on quantum Elman neural networks.

Research on wind field algorithm of wind lidar based on BP neural network and grey prediction

Science.gov (United States)

Chen, Yong; Chen, Chun-Li; Luo, Xiong; Zhang, Yan; Yang, Ze-hou; Zhou, Jie; Shi, Xiao-ding; Wang, Lei

2018-01-01

This paper uses the BP neural network and grey algorithm to forecast and study radar wind field. In order to reduce the residual error in the wind field prediction which uses BP neural network and grey algorithm, calculating the minimum value of residual error function, adopting the residuals of the gray algorithm trained by BP neural network, using the trained network model to forecast the residual sequence, using the predicted residual error sequence to modify the forecast sequence of the grey algorithm. The test data show that using the grey algorithm modified by BP neural network can effectively reduce the residual value and improve the prediction precision.

Neural networks for predicting breeding values and genetic gains

Directory of Open Access Journals (Sweden)

Gabi Nunes Silva

2014-12-01

Full Text Available Analysis using Artificial Neural Networks has been described as an approach in the decision-making process that, although incipient, has been reported as presenting high potential for use in animal and plant breeding. In this study, we introduce the procedure of using the expanded data set for training the network. Wealso proposed using statistical parameters to estimate the breeding value of genotypes in simulated scenarios, in addition to the mean phenotypic value in a feed-forward back propagation multilayer perceptron network. After evaluating artificial neural network configurations, our results showed its superiority to estimates based on linear models, as well as its applicability in the genetic value prediction process. The results further indicated the good generalization performance of the neural network model in several additional validation experiments.

The DSFPN, a new neural network for optical character recognition.

Science.gov (United States)

Morns, L P; Dlay, S S

1999-01-01

A new type of neural network for recognition tasks is presented in this paper. The network, called the dynamic supervised forward-propagation network (DSFPN), is based on the forward only version of the counterpropagation network (CPN). The DSFPN, trains using a supervised algorithm and can grow dynamically during training, allowing subclasses in the training data to be learnt in an unsupervised manner. It is shown to train in times comparable to the CPN while giving better classification accuracies than the popular backpropagation network. Both Fourier descriptors and wavelet descriptors are used for image preprocessing and the wavelets are proven to give a far better performance.

Separation prediction in two dimensional boundary layer flows using artificial neural networks

International Nuclear Information System (INIS)

Sabetghadam, F.; Ghomi, H.A.

2003-01-01

In this article, the ability of artificial neural networks in prediction of separation in steady two dimensional boundary layer flows is studied. Data for network training is extracted from numerical solution of an ODE obtained from Von Karman integral equation with approximate one parameter Pohlhousen velocity profile. As an appropriate neural network, a two layer radial basis generalized regression artificial neural network is used. The results shows good agreements between the overall behavior of the flow fields predicted by the artificial neural network and the actual flow fields for some cases. The method easily can be extended to unsteady separation and turbulent as well as compressible boundary layer flows. (author)

A mixed-scale dense convolutional neural network for image analysis

NARCIS (Netherlands)

D.M. Pelt (Daniël); J.A. Sethian (James)

2016-01-01

textabstractDeep convolutional neural networks have been successfully applied to many image-processing problems in recent works. Popular network architectures often add additional operations and connections to the standard architecture to enable training deeper networks. To achieve accurate results

Hybrid computing using a neural network with dynamic external memory.

Science.gov (United States)

Graves, Alex; Wayne, Greg; Reynolds, Malcolm; Harley, Tim; Danihelka, Ivo; Grabska-Barwińska, Agnieszka; Colmenarejo, Sergio Gómez; Grefenstette, Edward; Ramalho, Tiago; Agapiou, John; Badia, Adrià Puigdomènech; Hermann, Karl Moritz; Zwols, Yori; Ostrovski, Georg; Cain, Adam; King, Helen; Summerfield, Christopher; Blunsom, Phil; Kavukcuoglu, Koray; Hassabis, Demis

2016-10-27

Artificial neural networks are remarkably adept at sensory processing, sequence learning and reinforcement learning, but are limited in their ability to represent variables and data structures and to store data over long timescales, owing to the lack of an external memory. Here we introduce a machine learning model called a differentiable neural computer (DNC), which consists of a neural network that can read from and write to an external memory matrix, analogous to the random-access memory in a conventional computer. Like a conventional computer, it can use its memory to represent and manipulate complex data structures, but, like a neural network, it can learn to do so from data. When trained with supervised learning, we demonstrate that a DNC can successfully answer synthetic questions designed to emulate reasoning and inference problems in natural language. We show that it can learn tasks such as finding the shortest path between specified points and inferring the missing links in randomly generated graphs, and then generalize these tasks to specific graphs such as transport networks and family trees. When trained with reinforcement learning, a DNC can complete a moving blocks puzzle in which changing goals are specified by sequences of symbols. Taken together, our results demonstrate that DNCs have the capacity to solve complex, structured tasks that are inaccessible to neural networks without external read-write memory.

Tests of track segment and vertex finding with neural networks

International Nuclear Information System (INIS)

Denby, B.; Lessner, E.; Lindsey, C.S.

1990-04-01

Feed forward neural networks have been trained, using back-propagation, to find the slopes of simulated track segments in a straw chamber and to find the vertex of tracks from both simulated and real events in a more conventional drift chamber geometry. Network architectures, training, and performance are presented. 12 refs., 7 figs

Performance of an artificial neural network for vertical root fracture detection: an ex vivo study.

Science.gov (United States)

Kositbowornchai, Suwadee; Plermkamon, Supattra; Tangkosol, Tawan

2013-04-01

To develop an artificial neural network for vertical root fracture detection. A probabilistic neural network design was used to clarify whether a tooth root was sound or had a vertical root fracture. Two hundred images (50 sound and 150 vertical root fractures) derived from digital radiography--used to train and test the artificial neural network--were divided into three groups according to the number of training and test data sets: 80/120,105/95 and 130/70, respectively. Either training or tested data were evaluated using grey-scale data per line passing through the root. These data were normalized to reduce the grey-scale variance and fed as input data of the neural network. The variance of function in recognition data was calculated between 0 and 1 to select the best performance of neural network. The performance of the neural network was evaluated using a diagnostic test. After testing data under several variances of function, we found the highest sensitivity (98%), specificity (90.5%) and accuracy (95.7%) occurred in Group three, for which the variance of function in recognition data was between 0.025 and 0.005. The neural network designed in this study has sufficient sensitivity, specificity and accuracy to be a model for vertical root fracture detection. © 2012 John Wiley & Sons A/S.

Prediction of Aerodynamic Coefficient using Genetic Algorithm Optimized Neural Network for Sparse Data

Science.gov (United States)

Rajkumar, T.; Bardina, Jorge; Clancy, Daniel (Technical Monitor)

2002-01-01

Wind tunnels use scale models to characterize aerodynamic coefficients, Wind tunnel testing can be slow and costly due to high personnel overhead and intensive power utilization. Although manual curve fitting can be done, it is highly efficient to use a neural network to define the complex relationship between variables. Numerical simulation of complex vehicles on the wide range of conditions required for flight simulation requires static and dynamic data. Static data at low Mach numbers and angles of attack may be obtained with simpler Euler codes. Static data of stalled vehicles where zones of flow separation are usually present at higher angles of attack require Navier-Stokes simulations which are costly due to the large processing time required to attain convergence. Preliminary dynamic data may be obtained with simpler methods based on correlations and vortex methods; however, accurate prediction of the dynamic coefficients requires complex and costly numerical simulations. A reliable and fast method of predicting complex aerodynamic coefficients for flight simulation I'S presented using a neural network. The training data for the neural network are derived from numerical simulations and wind-tunnel experiments. The aerodynamic coefficients are modeled as functions of the flow characteristics and the control surfaces of the vehicle. The basic coefficients of lift, drag and pitching moment are expressed as functions of angles of attack and Mach number. The modeled and training aerodynamic coefficients show good agreement. This method shows excellent potential for rapid development of aerodynamic models for flight simulation. Genetic Algorithms (GA) are used to optimize a previously built Artificial Neural Network (ANN) that reliably predicts aerodynamic coefficients. Results indicate that the GA provided an efficient method of optimizing the ANN model to predict aerodynamic coefficients. The reliability of the ANN using the GA includes prediction of aerodynamic

PEAK TRACKING WITH A NEURAL NETWORK FOR SPECTRAL RECOGNITION

NARCIS (Netherlands)

COENEGRACHT, PMJ; METTING, HJ; VANLOO, EM; SNOEIJER, GJ; DOORNBOS, DA

1993-01-01

A peak tracking method based on a simulated feed-forward neural network with back-propagation is presented. The network uses the normalized UV spectra and peak areas measured in one chromatogram for peak recognition. It suffices to train the network with only one set of spectra recorded in one

Neural networks

International Nuclear Information System (INIS)

Denby, Bruce; Lindsey, Clark; Lyons, Louis

1992-01-01

The 1980s saw a tremendous renewal of interest in 'neural' information processing systems, or 'artificial neural networks', among computer scientists and computational biologists studying cognition. Since then, the growth of interest in neural networks in high energy physics, fueled by the need for new information processing technologies for the next generation of high energy proton colliders, can only be described as explosive

Causal Learning and Explanation of Deep Neural Networks via Autoencoded Activations

OpenAIRE

Harradon, Michael; Druce, Jeff; Ruttenberg, Brian

2018-01-01

Deep neural networks are complex and opaque. As they enter application in a variety of important and safety critical domains, users seek methods to explain their output predictions. We develop an approach to explaining deep neural networks by constructing causal models on salient concepts contained in a CNN. We develop methods to extract salient concepts throughout a target network by using autoencoders trained to extract human-understandable representations of network activations. We then bu...

Parameter estimation using compensatory neural networks

Indian Academy of Sciences (India)

of interconnections among neurons but also reduces the total computing time for training. The suggested model has properties of the basic neuron ..... Engelbrecht A P, Cloete I, Geldenhuys J, Zurada J M 1995 Automatic scaling using gamma learning for feedforward neural networks. From natural to artificial computing.

Particle identification using artificial neural networks at BESIII

International Nuclear Information System (INIS)

Qin Gang; Lv Junguang; Bian Jianming; Chinese Academy of Sciences, Beijing

2008-01-01

A multilayered perceptrons' neural network technique has been applied in the particle identification at BESIII. The networks are trained in each sub-detector level. The NN output of sub-detectors can be sent to a sequential network or be constructed as PDFs for a likelihood. Good muon-ID, electron-ID and hadron-ID are obtained from the networks by using the simulated Monte Carlo samples. (authors)

Development and Validation of a Deep Neural Network Model for Prediction of Postoperative In-hospital Mortality.

Science.gov (United States)

Lee, Christine K; Hofer, Ira; Gabel, Eilon; Baldi, Pierre; Cannesson, Maxime

2018-04-17

The authors tested the hypothesis that deep neural networks trained on intraoperative features can predict postoperative in-hospital mortality. The data used to train and validate the algorithm consists of 59,985 patients with 87 features extracted at the end of surgery. Feed-forward networks with a logistic output were trained using stochastic gradient descent with momentum. The deep neural networks were trained on 80% of the data, with 20% reserved for testing. The authors assessed improvement of the deep neural network by adding American Society of Anesthesiologists (ASA) Physical Status Classification and robustness of the deep neural network to a reduced feature set. The networks were then compared to ASA Physical Status, logistic regression, and other published clinical scores including the Surgical Apgar, Preoperative Score to Predict Postoperative Mortality, Risk Quantification Index, and the Risk Stratification Index. In-hospital mortality in the training and test sets were 0.81% and 0.73%. The deep neural network with a reduced feature set and ASA Physical Status classification had the highest area under the receiver operating characteristics curve, 0.91 (95% CI, 0.88 to 0.93). The highest logistic regression area under the curve was found with a reduced feature set and ASA Physical Status (0.90, 95% CI, 0.87 to 0.93). The Risk Stratification Index had the highest area under the receiver operating characteristics curve, at 0.97 (95% CI, 0.94 to 0.99). Deep neural networks can predict in-hospital mortality based on automatically extractable intraoperative data, but are not (yet) superior to existing methods.

Using neural networks to infer the hydrodynamic yield of aspherical sources

International Nuclear Information System (INIS)

Moran, B.; Glenn, L.

1993-01-01

We distinguish two kinds of difficulties with yield determination from aspherical sources. The first kind, the spoofing difficulty, occurs when a fraction of the energy of the explosion is channeled in such a way that it is not detected by the CORRTEX cable. In this case, neither neural networks nor any expert system can be expected to accurately estimate the yield without detailed information about device emplacement within the canister. Numerical simulations however, can provide an upper bound on the undetected fraction of the explosive energy. In the second instance, the interpretation difficulty, the data appear abnormal when analyzed using similar-explosion-scaling and the assumption of a spherical front. The inferred yield varies with time and the confidence in the yield estimate decreases. It is this kind of problem we address in this paper and for which neural networks can make a contribution. We used a back propagation neural network to infer the hydrodynamic yield of simulated aspherical sources. We trained the network using a subset of simulations from 3 different aspherical sources, with 3 different yield, and 3 satellite offset separations. The trained network was able to predict the yield within 15% in all cases and to identify the correct type of aspherical source in most cases. The predictive capability of the network increased with a larger training set. The neural network approach can easily incorporate information from new calculations or experiments and is therefore flexible and easy to maintain. We describe the potential capabilities and limitations in using such networks for yield estimations

Automatic Classification of volcano-seismic events based on Deep Neural Networks.

Science.gov (United States)

Titos Luzón, M.; Bueno Rodriguez, A.; Garcia Martinez, L.; Benitez, C.; Ibáñez, J. M.

2017-12-01

Seismic monitoring of active volcanoes is a popular remote sensing technique to detect seismic activity, often associated to energy exchanges between the volcano and the environment. As a result, seismographs register a wide range of volcano-seismic signals that reflect the nature and underlying physics of volcanic processes. Machine learning and signal processing techniques provide an appropriate framework to analyze such data. In this research, we propose a new classification framework for seismic events based on deep neural networks. Deep neural networks are composed by multiple processing layers, and can discover intrinsic patterns from the data itself. Internal parameters can be initialized using a greedy unsupervised pre-training stage, leading to an efficient training of fully connected architectures. We aim to determine the robustness of these architectures as classifiers of seven different types of seismic events recorded at "Volcán de Fuego" (Colima, Mexico). Two deep neural networks with different pre-training strategies are studied: stacked denoising autoencoder and deep belief networks. Results are compared to existing machine learning algorithms (SVM, Random Forest, Multilayer Perceptron). We used 5 LPC coefficients over three non-overlapping segments as training features in order to characterize temporal evolution, avoid redundancy and encode the signal, regardless of its duration. Experimental results show that deep architectures can classify seismic events with higher accuracy than classical algorithms, attaining up to 92% recognition accuracy. Pre-training initialization helps these models to detect events that occur simultaneously in time (such explosions and rockfalls), increase robustness against noisy inputs, and provide better generalization. These results demonstrate deep neural networks are robust classifiers, and can be deployed in real-environments to monitor the seismicity of restless volcanoes.

Neural network evaluation of tokamak current profiles for real time control

Science.gov (United States)

Wróblewski, Dariusz

1997-02-01

Active feedback control of the current profile, requiring real-time determination of the current profile parameters, is envisioned for tokamaks operating in enhanced confinement regimes. The distribution of toroidal current in a tokamak is now routinely evaluated based on external (magnetic probes, flux loops) and internal (motional Stark effect) measurements of the poloidal magnetic field. However, the analysis involves reconstruction of magnetohydrodynamic equilibrium and is too intensive computationally to be performed in real time. In the present study, a neural network is used to provide a mapping from the magnetic measurements (internal and external) to selected parameters of the safety factor profile. The single-pass, feedforward calculation of output of a trained neural network is very fast, making this approach particularly suitable for real-time applications. The network was trained on a large set of simulated equilibrium data for the DIII-D tokamak. The database encompasses a large variety of current profiles including the hollow current profiles important for reversed central shear operation. The parameters of safety factor profile (a quantity related to the current profile through the magnetic field tilt angle) estimated by the neural network include central safety factor, q0, minimum value of q, qmin, and the location of qmin. Very good performance of the trained neural network both for simulated test data and for experimental datais demonstrated.

Neural network evaluation of tokamak current profiles for real time control

International Nuclear Information System (INIS)

Wroblewski, D.

1997-01-01

Active feedback control of the current profile, requiring real-time determination of the current profile parameters, is envisioned for tokamaks operating in enhanced confinement regimes. The distribution of toroidal current in a tokamak is now routinely evaluated based on external (magnetic probes, flux loops) and internal (motional Stark effect) measurements of the poloidal magnetic field. However, the analysis involves reconstruction of magnetohydrodynamic equilibrium and is too intensive computationally to be performed in real time. In the present study, a neural network is used to provide a mapping from the magnetic measurements (internal and external) to selected parameters of the safety factor profile. The single-pass, feedforward calculation of output of a trained neural network is very fast, making this approach particularly suitable for real-time applications. The network was trained on a large set of simulated equilibrium data for the DIII-D tokamak. The database encompasses a large variety of current profiles including the hollow current profiles important for reversed central shear operation. The parameters of safety factor profile (a quantity related to the current profile through the magnetic field tilt angle) estimated by the neural network include central safety factor, q 0 , minimum value of q, q min , and the location of q min . Very good performance of the trained neural network both for simulated test data and for experimental datais demonstrated. copyright 1997 American Institute of Physics

A novel and generalized approach in the inversion of geoelectrical resistivity data using Artificial Neural Networks (ANN)

Science.gov (United States)

Raj, A. Stanley; Srinivas, Y.; Oliver, D. Hudson; Muthuraj, D.

2014-03-01

The non-linear apparent resistivity problem in the subsurface study of the earth takes into account the model parameters in terms of resistivity and thickness of individual subsurface layers using the trained synthetic data by means of Artificial Neural Networks (ANN). Here we used a single layer feed-forward neural network with fast back propagation learning algorithm. So on proper training of back propagation networks it tends to give the resistivity and thickness of the subsurface layer model of the field resistivity data with reference to the synthetic data trained in the appropriate network. During training, the weights and biases of the network are iteratively adjusted to make network performance function level more efficient. On adequate training, errors are minimized and the best result is obtained using the artificial neural networks. The network is trained with more number of VES data and this trained network is demonstrated by the field data. The accuracy of inversion depends upon the number of data trained. In this novel and specially designed algorithm, the interpretation of the vertical electrical sounding has been done successfully with the more accurate layer model.

Target recognition based on convolutional neural network

Science.gov (United States)

Wang, Liqiang; Wang, Xin; Xi, Fubiao; Dong, Jian

2017-11-01

One of the important part of object target recognition is the feature extraction, which can be classified into feature extraction and automatic feature extraction. The traditional neural network is one of the automatic feature extraction methods, while it causes high possibility of over-fitting due to the global connection. The deep learning algorithm used in this paper is a hierarchical automatic feature extraction method, trained with the layer-by-layer convolutional neural network (CNN), which can extract the features from lower layers to higher layers. The features are more discriminative and it is beneficial to the object target recognition.

Human Face Recognition Using Convolutional Neural Networks

Directory of Open Access Journals (Sweden)

Răzvan-Daniel Albu

2009-10-01

Full Text Available In this paper, I present a novel hybrid face recognition approach based on a convolutional neural architecture, designed to robustly detect highly variable face patterns. The convolutional network extracts successively larger features in a hierarchical set of layers. With the weights of the trained neural networks there are created kernel windows used for feature extraction in a 3-stage algorithm. I present experimental results illustrating the efficiency of the proposed approach. I use a database of 796 images of 159 individuals from Reims University which contains quite a high degree of variability in expression, pose, and facial details.

Modelling and Prediction of Photovoltaic Power Output Using Artificial Neural Networks

Directory of Open Access Journals (Sweden)

Aminmohammad Saberian

2014-01-01

Full Text Available This paper presents a solar power modelling method using artificial neural networks (ANNs. Two neural network structures, namely, general regression neural network (GRNN feedforward back propagation (FFBP, have been used to model a photovoltaic panel output power and approximate the generated power. Both neural networks have four inputs and one output. The inputs are maximum temperature, minimum temperature, mean temperature, and irradiance; the output is the power. The data used in this paper started from January 1, 2006, until December 31, 2010. The five years of data were split into two parts: 2006–2008 and 2009-2010; the first part was used for training and the second part was used for testing the neural networks. A mathematical equation is used to estimate the generated power. At the end, both of these networks have shown good modelling performance; however, FFBP has shown a better performance comparing with GRNN.

Applying neural networks to control the TFTR neutral beam ion sources

International Nuclear Information System (INIS)

Lagin, L.

1992-01-01

This paper describes the application of neural networks to the control of the neutral beam long-pulse positive ion source accelerators on the Tokamak Fusion Test Reactor (TFTR) at Princeton University. Neural networks were used to learn how the operators adjust the control setpoints when running these sources. The data sets used to train these networks were derived from a large database containing actual setpoints and power supply waveform calculations for the 1990 run period. The networks learned what the optimum control setpoints should initially be set based uon desired accel voltage and perveance levels. Neural networks were also used to predict the divergence of the ion beam

Maximum solid concentrations of coal water slurries predicted by neural network models

Energy Technology Data Exchange (ETDEWEB)

Cheng, Jun; Li, Yanchang; Zhou, Junhu; Liu, Jianzhong; Cen, Kefa

2010-12-15

The nonlinear back-propagation (BP) neural network models were developed to predict the maximum solid concentration of coal water slurry (CWS) which is a substitute for oil fuel, based on physicochemical properties of 37 typical Chinese coals. The Levenberg-Marquardt algorithm was used to train five BP neural network models with different input factors. The data pretreatment method, learning rate and hidden neuron number were optimized by training models. It is found that the Hardgrove grindability index (HGI), moisture and coalification degree of parent coal are 3 indispensable factors for the prediction of CWS maximum solid concentration. Each BP neural network model gives a more accurate prediction result than the traditional polynomial regression equation. The BP neural network model with 3 input factors of HGI, moisture and oxygen/carbon ratio gives the smallest mean absolute error of 0.40%, which is much lower than that of 1.15% given by the traditional polynomial regression equation. (author)