- Home
- ▪
- About
- ▪
- News
- ▪
- Advanced Search
- ▪
- Mobile
- ▪
- Contact Us
- ▪
- Site Map
- ▪
- Help

1

Modular, Multilayer Perceptron

Combination of proposed modular, multilayer perceptron and algorithm for its operation recognizes new objects after relatively brief retraining sessions. (Perceptron is multilayer, feedforward artificial neural network fully connected and trained via back-propagation learning algorithm.) Knowledge pertaining to each object to be recognized resides in subnetwork of full network, therefore not necessary to retrain full network to recognize each new object.

Cheng, Li-Jen; Liu, Tsuen-Hsi

1991-01-01

2

Prediction of Parametric Roll Resonance by Multilayer Perceptron Neural Network

DEFF Research Database (Denmark)

Parametric roll resonance is a ship stability related phenomenon that generates sudden large amplitude oscillations up to 30-40 degrees of roll. This can cause severe damage, and it can put the crew in serious danger. The need for a parametric rolling real time prediction system has been acknowledged in the last few years. This work proposes a prediction system based on a multilayer perceptron (MP) neural network. The training and testing of the MP network is accomplished by feeding it with simulated data of a three degrees-of-freedom nonlinear model of a fishing vessel. The neural network is shown to be capable of forecasting the ship’s roll motion in realistic scenarios.

Galeazzi, Roberto; Blanke, Mogens

2011-01-01

3

Inversion of Self Potential Anomalies with Multilayer Perceptron Neural Networks

This study investigates the inverse solution on a buried and polarized sphere-shaped body using the self-potential method via multilayer perceptron neural networks (MLPNN). The polarization angle ( ?), depth to the centre of sphere ( h), electrical dipole moment ( K) and the zero distance from the origin ( x 0) were estimated. For testing the success of the MLPNN for sphere model, parameters were also estimated by the traditional Damped Least Squares (Levenberg-Marquardt) inversion technique (DLS). The MLPNN was first tested on a synthetic example. The performance of method was also tested for two S/N ratios (5 % and 10 %) by adding noise to the same synthetic data, the estimated model parameters with MLPNN and DLS method are satisfactory. The MLPNN also applied for the field data example in ?zmir, Urla district, Turkey, with two cross-section data evaluated by MLPNN and DLS, and the two methods showed good agreement.

Kaftan, Ilknur; S?nd?rg?, Petek; Akdemir, Özer

2014-08-01

4

Classification of fused face images using multilayer perceptron neural network

This paper presents a concept of image pixel fusion of visual and thermal faces, which can significantly improve the overall performance of a face recognition system. Several factors affect face recognition performance including pose variations, facial expression changes, occlusions, and most importantly illumination changes. So, image pixel fusion of thermal and visual images is a solution to overcome the drawbacks present in the individual thermal and visual face images. Fused images are projected into eigenspace and finally classified using a multi-layer perceptron. In the experiments we have used Object Tracking and Classification Beyond Visible Spectrum (OTCBVS) database benchmark thermal and visual face images. Experimental results show that the proposed approach significantly improves the verification and identification performance and the success rate is 95.07%. The main objective of employing fusion is to produce a fused image that provides the most detailed and reliable information. Fusion of multip...

Bhattacharjee, Debotosh; Nasipuri, Mita; Basu, Dipak Kumar; Kundu, Mahantapas

2010-01-01

5

Directory of Open Access Journals (Sweden)

Full Text Available The Multi-layer Perceptron Neural Networks (MLP NN are well known for their simplicity, ease of training for small-scale problems, and suitability for online implementation. This paper presents the methodology and challenges in the design of near-optimal MLP NN based classifier with maximize classification accuracy under the constraints of minimum network dimension for implementation intelligent sensors.

Nadir N. Charniya

2013-02-01

6

Directory of Open Access Journals (Sweden)

Full Text Available The purpose of the present research is to apply a Multilayer Perceptron (MLP neural network technique to create classification models from a portfolio of Non-Performing Loans (NPLs to classify this type of credit derivative. These credit derivatives are characterized as the amount of loans that were not paid and are already overdue more than 90 days. Since these titles are, because of legislative motives, moved by losses, Credit Rights Investment Funds (FDIC performs the purchase of these debts and the recovery of the credits. Using the Multilayer Perceptron (MLP architecture of Artificial Neural Network (ANN, classification models regarding the posterior recovery of these debts were created. To evaluate the performance of the models, evaluation metrics of classification relating to the neural networks with different architectures were presented. The results of the classifications were satisfactory, given the classification models were successful in the presented economics costs structure.

Flávio Clésio Silva de Souza

2014-06-01

7

Apply Multi-Layer Perceptrons Neural Network for Off-Line Signature Verification and Recognition

Directory of Open Access Journals (Sweden)

Full Text Available This paper discusses the applying of Multi-layer perceptrons for signature verification and recognition using a new approach enables the user to recognize whether a signature is original or a fraud. The approach starts by scanning images into the computer, then modifying their quality through image enhancement and noise reduction, followed by feature extraction and neural network training, and finally verifies the authenticity of the signature. The paper discusses the different stages of the process including: image pre-processing, feature extraction and pattern recognition through neural networks.

Suhail Odeh

2011-11-01

8

Neural and stochastic models for signal classification generate output probabilities to indicate whether or not their inputs are members of the modeled class. This paper presents a feature enhancing neural network with weights based on the modeled class which can improve the classification performance of single output classifiers, by increasing output probabilities for members of the modeled class or decreasing output probabilities for non-members. The neural network is demonstrated as a front-end for multi-layer perceptron and semi-continuous hidden Markov model based classifiers for speech recognition applications. It is unique in that the weights and width of the input layer adapt based on extracted characteristics from the input speech signal. The connectionist architecture is motivated by the highly successful time-delay neural network and the desire to find efficient training procedures for class-dependent, short- time transformations. The weights are determined using a principal component analysis process and can be found by applying iterative or conventional algorithms. The neural network reduces false acceptances by more than one-third for a defined mono-syllable keyword spotting application using a semi-continuous hidden Markov model based system. An evaluation of the neural network as a front-end for multi-layer perceptron based classifiers which distinguish a word from confusable words is also presented.

Clary, Gregory J.; Hansen, John H. L.

1992-12-01

9

Classification of fuels using multilayer perceptron neural networks

International Nuclear Information System (INIS)

Electrical impedance data obtained with an array of conducting polymer chemical sensors was used by a neural network (ANN) to classify fuel adulteration. Real samples were classified with accuracy greater than 90% in two groups: approved and adulterated.

10

Highly Accurate Multi-layer Perceptron Neural Network for Air Data System

Directory of Open Access Journals (Sweden)

Full Text Available The error backpropagation multi-layer perceptron algorithm is revisited. This algorithm is used to train and validate two models of three-layer neural networks that can be used to calibrate a 5-hole pressure probe. This paper addresses Occam.s Razor problem as it describes the adhoc training methodology applied to improve accuracy and sensitivity. The trained outputs from 5-4-3 feed-forward network architecture with jump connection are comparable to second decimal digit (~0.05 accuracy, hitherto unreported in literature.

H.S. Krishna

2009-11-01

11

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, the problem of predicting the exchange rate time series in the foreign exchange rate market is going to be solved using a time-delayed multilayer perceptron neural network with gold price as external factor. The input for the learning phase of the artificial neural network are the exchange rate data of the last five days plus the gold price in two different currencies of the exchange rate as the external factor for helping the artificial neural network improving its forecast accuracy. The five-day delay has been chosen because of the weekly cyclic behavior of the exchange rate time series with the consideration of two holidays in a week. The result of forecasts are then compared with using the multilayer peceptron neural network without gold price external factor by two most important evaluation techniques in the literature of exchange rate prediction. For the experimental analysis phase, the data of three important exchange rates of EUR/USD, GBP/USD, and USD/JPY are used.

Mohammad Fathian

2012-04-01

12

Analysis of (7)Be behaviour in the air by using a multilayer perceptron neural network.

A multilayer perceptron artificial neural network (ANN) model for the prediction of the (7)Be behaviour in the air as the function of meteorological parameters was developed. The model was optimized and tested using (7)Be activity concentrations obtained by standard gamma-ray spectrometric analysis of air samples collected in Belgrade (Serbia) during 2009-2011 and meteorological data for the same period. Good correlation (r = 0.91) between experimental values of (7)Be activity concentrations and those predicted by ANN was obtained. The good performance of the model in prediction of (7)Be activity concentrations could provide basis for construction of models which would forecast behaviour of other airborne radionuclides. PMID:25106024

Samolov, A; Dragovi?, S; Dakovi?, M; Ba?i?, G

2014-11-01

13

Near-infrared (NIR) spectroscopy is being applied to the solution of problems in many areas of biomedical and pharmaceutical research. In this paper we investigate the use of NIR spectroscopy as an analytical tool to quantify concentrations of urea, creatinine, glucose and oxyhemoglobin (HbO2). Measurements have been made in vitro with a portable spectrometer developed in our labs that consists of a two beam interferometer operating in the range of 800-2300 nm. For the data analysis a pattern recognition philosophy was used with a preprocessing stage and a multi-layer perceptron (MLP) neural network for the measurement stage. Results show that the interferogram signatures of the above compounds are sufficiently strong in that spectral range. Measurements of three different concentrations were possible with mean squared error (MSE) of the order of 10(-6). PMID:17947035

Kalamatianos, Dimitrios; Liatsis, Panos; Wellstead, Peter E

2006-01-01

14

Geomagnetic Dst index forecast using a multilayer perceptrons artificial neural network

International Nuclear Information System (INIS)

Complete text of publication follows. The best known manifestations of the impact of solar wind on the magnetosphere are the geomagnetic storms. The prediction of geomagnetic field behavior allows the alert of geomagnetic storms occurrence, as those phenomena can cause many damages in the planet. The Artificial Intelligence tools have been applied in many multidisciplinary studies, covering several areas of knowledge, as a choice of approach to the solution of problems with characteristics like non-linearity, imprecision, and other features that can not be easily solved with conventional computational models. Techniques such as Artificial Neural Networks, Expert Systems and Decision Trees have been used in the Space Weather studies to perform tasks such as forecasting geomagnetic storms and the investigation of rules and parameters related on its occurrence. The main focus of this work is on forecasting the geomagnetic field behavior, represented this time by the Dst index, using for that task, mainly, the interplanetary magnetic field components and solar wind data. The tool chosen here to solve the non-linear problem was a Multi-layer Perceptrons Artificial Neural Network, trained with the backpropagation algorithm. Unlike what was done in other studies, we chose to predict calm and disturbed periods like, for example, a full month of data, for application in a real time forecasting system. It was possible to predict the geomagnetic Dst index one or two hours before with great percentage efficiency.

15

In the past decade, the treatment amount of municipal solid waste (MSW) by incineration has increased significantly in Taiwan. By year 2008, approximately 70% of the total MSW generated will be incinerated. The energy content (usually expressed by lower heating value [LHV]) of MSW is an important parameter for the selection of incinerator capacity. In this work, wastes from 55 sampling sites, including villages, towns, cities, and remote islands in the Taiwan area, were sampled and analyzed once a season from April 2002 to March 2003 to determine the waste characteristics. The LHV of MSW in Taiwan was predicted by the multilayer perceptron (MLP) neural networks model using the input parameters of elemental analysis and dry- or wet-base physical compositions. Although all three of the models predicted LHV values rather accurately, the elemental analysis model provided the most accurate prediction of LHV values. Additionally, the wet-base physical composition model was the easiest and most economical. Therefore, the waste treatment operators can choose the more appropriate analysis method considering situations themselves, such as time, equipment, technology, and cost. PMID:16805410

Shu, Hung-Yee; Lu, Hsin-Chung; Fan, Huan-Jung; Chang, Ming-Chin; Chen, Jyh-Cherng

2006-06-01

16

Cross Validation Evaluation for Breast Cancer Prediction Using Multilayer Perceptron Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available Problem statement: The presence of metastasis in the regional lymph nodes is the most important factor in predicting prognosis in breast cancer. Many biomarkers have been identified that appear to relate to the aggressive behaviour of cancer. However, the nonlinear relation of these markers to nodal status and also the existence of complex interaction between markers have prohibited an accurate prognosis. Approach: The aim of this study is to investigate the effectiveness of a Multilayer Perceptron (MLP for predicting breast cancer progression using a set of four biomarkers of breast tumors. The biomarkers include DNA ploidy, cell cycle distribution (G0G1/G2M, steroid receptors (ER/PR and S-Phase Fraction (SPF. A further objective of the study is to explore the predictive potential of these markers in defining the state of nodal involvement in breast cancer. Two methods of outcome evaluation viz. stratified and simple k-fold Cross Validation (CV are studied in order to assess their accuracy and reliability for neural network validation. Criteria such as output accuracy, sensitivity and specificity are used for selecting the best validation technique besides evaluating the network outcome for different combinations of markers. Results: The results show that stratified 2-fold CV is more accurate and reliable compared to simple k-fold CV as it obtains a higher accuracy and specificity and also provides a more stable network validation in terms of sensitivity. Best prediction results are obtained by using an individual marker-SPF which obtains an accuracy of 65%. Conclusion/Recommendations: Our findings suggest that MLP-based analysis provides an accurate and reliable platform for breast cancer prediction given that an appropriate design and validation method is employed.

Shirin A. Mojarad

2011-01-01

17

Multilayered perceptron neural networks to compute energy losses in magnetic cores

International Nuclear Information System (INIS)

This paper presents a new approach based on multilayered perceptrons (MLPs) to compute the specific energy losses of toroidal wound cores built from 3% SiFe 0.27 mm thick M4, 0.1 and 0.08 mm thin gauge electrical steel strips. The MLP has been trained by a back-propagation and extended delta-bar-delta learning algorithm. The results obtained by using the MLP model were compared with a commonly used conventional method. The comparison has shown that the proposed model improved loss estimation with respect to the conventional method

18

Quaternionic Multilayer Perceptron with Local Analyticity

Directory of Open Access Journals (Sweden)

Full Text Available A multi-layered perceptron type neural network is presented and analyzed in this paper. All neuronal parameters such as input, output, action potential and connection weight are encoded by quaternions, which are a class of hypercomplex number system. Local analytic condition is imposed on the activation function in updating neurons’ states in order to construct learning algorithm for this network. An error back-propagation algorithm is introduced for modifying the connection weights of the network.

Nobuyuki Matsui

2012-11-01

19

Digital Repository Infrastructure Vision for European Research (DRIVER)

Artificial neural networks (ANNs) are comparatively straightforward to understand and use in the analysis of scientific data. However, this relative transparency may encourage their use in an uncritical, and therefore possibly unproductive, fashion. The geometry of a network is among the most crucial factors in the successful deployment of network tools; in this review, we cover methods that can be used to determine optimum or near-optimum geometries. These methods of determining neural netwo...

Curteanu, S.; Cartwright, H.

2011-01-01

20

International Nuclear Information System (INIS)

Critical heat flux (CHF) is an important parameter for the design of nuclear reactors. Although many experimental and theoretical researches have been performed, there is not a single correlation to predict CHF because it is influenced by many parameters. These parameters are based on fixed inlet, local and fixed outlet conditions. Artificial neural networks (ANNs) have been applied to a wide variety of different areas such as prediction, approximation, modeling and classification. In this study, two types of neural networks, radial basis function (RBF) and multilayer perceptron (MLP), are trained with the experimental CHF data and their performances are compared. RBF predicts CHF with root mean square (RMS) errors of 0.24%, 7.9%, 0.16% and MLP predicts CHF with RMS errors of 1.29%, 8.31% and 2.71%, in fixed inlet conditions, local conditions and fixed outlet conditions, respectively. The results show that neural networks with RBF structure have superior performance in CHF data prediction over MLP neural networks. The parametric trends of CHF obtained by the trained ANNs are also evaluated and results reported

21

Directory of Open Access Journals (Sweden)

Full Text Available The prediction of temporal concentration profiles of a transported pollutant in a river is still a subject of ongoing research efforts worldwide. The present paper is aimed at studying the possibility of using Multi-Layer Perceptron Neural Networks to evaluate the whole concentration versus time profile at several cross-sections of a river under various flow conditions, using as little information about the river system as possible. In contrast with the earlier neural networks based work on longitudinal dispersion coefficients, this new approach relies more heavily on measurements of concentration collected during tracer tests over a range of flow conditions, but fewer hydraulic and morphological data are needed. The study is based upon 26 tracer experiments performed in a small river in Edinburgh, UK (Murray Burn at various flow rates in a 540 m long reach. The only data used in this study were concentration measurements collected at 4 cross-sections, distances between the cross-sections and the injection site, time, as well as flow rate and water velocity, obtained according to the data measured at the 1st and 2nd cross-sections.

The four main features of concentration versus time profiles at a particular cross-section, namely the peak concentration, the arrival time of the peak at the cross-section, and the shapes of the rising and falling limbs of the profile are modeled, and for each of them a separately designed neural network was used. There was also a variant investigated in which the conservation of the injected mass was assured by adjusting the predicted peak concentration. The neural network methods were compared with the unit peak attenuation curve concept.

In general the neural networks predicted the main features of the concentration profiles satisfactorily. The predicted peak concentrations were generally better than those obtained using the unit peak attenuation method, and the method with mass-conservation assured generally performed better than the method that did not account for mass-conservation. Predictions of peak travel time were also better using the neural networks than the unit peak attenuation method. Including more data into the neural network training set clearly improved the prediction of the shapes of the concentration profiles. Similar improvements in peak concentration were less significant and the travel time prediction appeared to be largely unaffected.

A. Piotrowski

2007-12-01

22

Directory of Open Access Journals (Sweden)

Full Text Available The prediction of temporal concentration profiles of a transported pollutant in a river is still a subject of ongoing research efforts worldwide. The present paper is aimed at studying the possibility of using Multi-Layer Perceptron Neural Networks to evaluate the whole concentration versus time profile at several cross-sections of a river under various flow conditions, using as little information about the river system as possible. In contrast with the earlier neural networks based work on longitudinal dispersion coefficients, this new approach relies more heavily on measurements of concentration collected during tracer tests over a range of flow conditions, but fewer hydraulic and morphological data are needed. The study is based upon 26 tracer experiments performed in a small river in Edinburgh, UK (Murray Burn at various flow rates in a 540 m long reach. The only data used in this study were concentration measurements collected at 4 cross-sections, distances between the cross-sections and the injection site, time, as well as flow rate and water velocity, obtained according to the data measured at the 1st and 2nd cross-sections.

The four main features of concentration versus time profiles at a particular cross-section, namely the peak concentration, the arrival time of the peak at the cross-section, and the shapes of the rising and falling limbs of the profile are modeled, and for each of them a separately designed neural network was used. There was also a variant investigated in which the conservation of the injected mass was assured by adjusting the predicted peak concentration. The neural network methods were compared with the unit peak attenuation curve concept.

In general the neural networks predicted the main features of the concentration profiles satisfactorily. The predicted peak concentrations were generally better than those obtained using the unit peak attenuation method, and the method with mass-conservation assured generally performed better than the method that did not account for mass-conservation. Predictions of peak travel time were also better using the neural networks than the unit peak attenuation method. Including more data into the neural network training set clearly improved the prediction of the shapes of the concentration profiles. Similar improvements in peak concentration were less significant and the travel time prediction appeared to be largely unaffected.

A. Piotrowski

2007-08-01

23

Multilayer Perceptrons to Approximate Quaternion Valued Functions.

In this paper a new type of multilayer feedforward neural network is introduced. Such a structure, called hypercomplex multilayer perceptron (HMLP), is developed in quaternion algebra and allows quaternionic input and output signals to be dealt with, requiring a lower number of neurons than the real MLP, thus providing a reduced computational complexity. The structure introduced represents a generalization of the multilayer perceptron in the complex space (CMLP) reported in the literature. The fundamental result reported in the paper is a new density theorem which makes HMLPs universal interpolators of quaternion valued continuous functions. Moreover the proof of the density theorem can be restricted in order to formulate a density theorem in the complex space. Due to the identity between the quaternion and the four-dimensional real space, such a structure is also useful to approximate multidimensional real valued functions with a lower number of real parameters, decreasing the probability of being trapped in local minima during the learning phase. A numerical example is also reported in order to show the efficiency of the proposed structure. Copyright 1997 Elsevier Science Ltd. All Rights Reserved. PMID:12662531

Xibilia, M G.; Muscato, G; Fortuna, L; Arena, P

1997-03-01

24

Self-potential (SP) is one of the oldest geophysical methods that provides important information about near-surface structures. Several methods have been developed to interpret SP data using simple geometries. This study investigated inverse solution of a buried, polarized sphere-shaped self-potential (SP ) anomaly via Multilayer Perceptron Neural Networks ( MLPNN ). The polarization angle ( ? ) and depth to the centre of sphere ( h )were estimated. The MLPNN is applied to synthetic and field SP data. In order to see the capability of the method in detecting the number of sources, MLPNN was applied to different spherical models at different depths and locations.. Additionally, the performance of MLPNN was tested by adding random noise to the same synthetic test data. The sphere model successfully obtained similar parameters under different S/N ratios. Then, MLPNN method was applied to two field examples. The first one is the cross section taken from the SP anomaly map of the Ergani-Süleymanköy (Turkey) copper mine. MLPNN was also applied to SP data from Seferihisar Izmir (Western Turkey) geothermal field. The MLPNN results showed good agreement with the original synthetic data set. The effect of The technique gave satisfactory results following the addition of 5% and 10% Gaussian noise levels. The MLPNN results were compared to other SP interpretation techniques, such as Normalized Full Gradient (NFG), inverse solution and nomogram methods. All of the techniques showed strong similarity. Consequently, the synthetic and field applications of this study show that MLPNN provides reliable evaluation of the self potential data modelled by the sphere model.

Kaftan, Ilknur; Sindirgi, Petek

2013-04-01

25

Auto-kernel using multilayer perceptron

Directory of Open Access Journals (Sweden)

Full Text Available This work presents a constructive method to train the multilayer perceptron layer after layer successively and to accomplish the kernel used in the support vector machine. Data in different classes will be trained to map to distant points in each layer. This will ease the mapping of the next layer. A perfect mapping kernel can be accomplished successively. Those distant mapped points can be discriminated easily by a single perceptron.

Wei-Chen Cheng

2012-06-01

26

Fourier-Lapped Multilayer Perceptron Method for Speech Quality Assessment

Directory of Open Access Journals (Sweden)

Full Text Available The paper introduces a new objective method for speech quality assessment called Fourier-lapped multilayer perceptron (FLMLP. This method uses an overcomplete transform based on the discrete Fourier transform (DFT and modulated lapped transform (MLT. This transform generates the DFT and the MLT speech spectral domains from which several relevant perceptual parameters are extracted. The proposed method also employs a multilayer perceptron neural network trained by a modified version of the scaled conjugated gradient method. This neural network maps the perceptual parameters into a subjective score. The numerical results show that FLMLP is an effective alternative to previous methods. As a result, it is worth stating that the techniques here described may be potentially useful to other researches facing the same kind of problem.

Ribeiro MoisésVidal

2005-01-01

27

A Parallel Framework for Multilayer Perceptron for Human Face Recognition

Digital Repository Infrastructure Vision for European Research (DRIVER)

Artificial neural networks have already shown their success in face recognition and similar complex pattern recognition tasks. However, a major disadvantage of the technique is that it is extremely slow during training for larger classes and hence not suitable for real-time complex problems such as pattern recognition. This is an attempt to develop a parallel framework for the training algorithm of a perceptron. In this paper, two general architectures for a Multilayer Perce...

Mita Nasipuri; Mahantapas Kundu; Dipak Kumar Basu; Debotosh Bhattacharjee; Mrinal Kanti Bhowmik

2010-01-01

28

A Parallel Framework for Multilayer Perceptron for Human Face Recognition

Artificial neural networks have already shown their success in face recognition and similar complex pattern recognition tasks. However, a major disadvantage of the technique is that it is extremely slow during training for larger classes and hence not suitable for real-time complex problems such as pattern recognition. This is an attempt to develop a parallel framework for the training algorithm of a perceptron. In this paper, two general architectures for a Multilayer Perceptron (MLP) have been demonstrated. The first architecture is All-Class-in-One-Network (ACON) where all the classes are placed in a single network and the second one is One-Class-in-One-Network (OCON) where an individual single network is responsible for each and every class. Capabilities of these two architectures were compared and verified in solving human face recognition, which is a complex pattern recognition task where several factors affect the recognition performance like pose variations, facial expression changes, occlusions, and ...

Bhowmik, M K; Nasipuri, M; Basu, D K; Kundu, M

2010-01-01

29

This study proposes the use of multi-layer perceptron neural networks (MLPNN) to invert dispersion curves obtained via multi-channel analysis of surface waves (MASW) for shear S-wave velocity profile. The dispersion curve used in inversion includes the fundamental-mode dispersion data. In order to investigate the applicability and performance of the proposed MLPNN algorithm, test studies were performed using both synthetic and field examples. Gaussian random noise with a standard deviation of 4 and 8% was added to the noise-free test data to make the synthetic test more realistic. The model parameters, such as S-wave velocities and thicknesses of the synthetic layered-earth model, were obtained for different S/N ratios and noise-free data. The field survey was performed over the natural gas pipeline, located in the Germencik district of Ayd?n city, western Turkey. The results show that depth, velocity, and location of the embedded natural gas pipe are successfully estimated with reasonably good approximation.

Çaylak, Ça?r?; Kaftan, ?lknur

2014-04-01

30

International Nuclear Information System (INIS)

In this work, multilayered perceptron neural networks (MLPNNs) were presented for the computation of the gamma-ray energy absorption buildup factors (BA) of seven thermoluminescent dosimetric (TLD) materials [LiF, BeO, Na2B4O7, CaSO4, Li2B4O7, KMgF3, Ca3(PO4)2] in the energy region 0.015–15 MeV, and for penetration depths up to 10 mfp (mean-free-path). The MLPNNs have been trained by a Levenberg–Marquardt learning algorithm. The developed model is in 99% agreement with the ANSI/ANS-6.4.3 standard data set. Furthermore, the model is fast and does not require tremendous computational efforts. The estimated BA data for TLD materials have been given with penetration depth and incident photon energy as comparative to the results of the interpolation method using the Geometrical Progression (G-P) fitting formula. - Highlights: ? Gamma-ray energy absorption buildup factors estimation in TLD materials. ? The ANN approach can be alternative to G-P fitting method for BA calculations. ? The applied model is not time-consuming and easily predicted

31

Directory of Open Access Journals (Sweden)

Full Text Available There are several neural network implementations using either software, hardware-based or a hardware/software co-design. This work proposes a hardware architecture to implement an artificial neural network (ANN, whose topology is the multilayer perceptron (MLP. In this paper, we explore the parallelism of neural networks and allow on-thefly changes of the number of inputs, number of layers and number of neurons per layer of the net. This reconfigurability characteristic permits that any application of ANNs may be implemented using the proposed hardware. In order to reduce the processing time that is spent in arithmetic computation, a real number is represented using a fraction of integers. In this way, the arithmetics is limited to integer operations, performed by fast combinational circuits. A simple state machine is required to control sums and products of fractions. Sigmoid is used as the activation function in the proposed implementation. It is approximated by polynomials, whose underlying computation requires only sums and products. A theorem is introduced and proven so as to cover the arithmetic strategy of the computation of the activation function. Thus, the arithmetic circuitry used to implement the neuron weighted sum is reused for computing the sigmoid. this resource sharing decreased drastically the total area of the system. After modeling and simulation for functionality validation, the proposed architecture synthesized using reconfigurable hardware. The results are promising.Em termos computacionais, uma rede neural artificial (RNA pode ser implementada em software ou em hardware, ou ainda de maneira híbrida, combinando ambos os recursos. O presente trabalho propõe uma arquitetura de hardware para a computação de uma rede neural do tipo perceptron com múltiplas camadas (MLP. Soluções em hardware tendem a ser mais eficientes do que soluções em software. O projeto em questão, além de explorar fortemente o paralelismo das redes neurais, permite alterações do número de entradas, número de camadas e de neurônios por camada, de modo que diversas aplicações de RNAs possam ser executadas no hardware proposto. Visando a uma redução de tempo do processamento aritmético, um número real é aproximado por uma fração de inteiros. Dessa forma, as operações aritméticas limitam-se a operações inteiras, executadas por circuitos combinacionais. Uma simples máquina de estados é demandada para controlar somas e produtos de frações. A função de ativação usada neste projeto é a sigmóide. Essa função é aproximada mediante o uso de polinômios, cujas operações são regidas por somas e produtos. Um teorema é introduzido e provado, permitindo a fundamentação da estratégia de cálculo da função de ativação. Dessa forma, reaproveita-se o circuito aritmético da soma ponderada para também computar a sigmóide. Essa re-utilização dos recursos levou a uma redução drástica de área total de circuito. Após modelagem e simulação para validação do bom funcionamento, a arquitetura proposta foi sintetizada utilizando recursos reconfiguráveis, do tipo FPGA. Os resultados são promissores.

Rodrigo Martins da Silva

2011-12-01

32

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: English Abstract in portuguese Em termos computacionais, uma rede neural artificial (RNA) pode ser implementada em software ou em hardware, ou ainda de maneira híbrida, combinando ambos os recursos. O presente trabalho propõe uma arquitetura de hardware para a computação de uma rede neural do tipo perceptron com múltiplas camadas [...] (MLP). Soluções em hardware tendem a ser mais eficientes do que soluções em software. O projeto em questão, além de explorar fortemente o paralelismo das redes neurais, permite alterações do número de entradas, número de camadas e de neurônios por camada, de modo que diversas aplicações de RNAs possam ser executadas no hardware proposto. Visando a uma redução de tempo do processamento aritmético, um número real é aproximado por uma fração de inteiros. Dessa forma, as operações aritméticas limitam-se a operações inteiras, executadas por circuitos combinacionais. Uma simples máquina de estados é demandada para controlar somas e produtos de frações. A função de ativação usada neste projeto é a sigmóide. Essa função é aproximada mediante o uso de polinômios, cujas operações são regidas por somas e produtos. Um teorema é introduzido e provado, permitindo a fundamentação da estratégia de cálculo da função de ativação. Dessa forma, reaproveita-se o circuito aritmético da soma ponderada para também computar a sigmóide. Essa re-utilização dos recursos levou a uma redução drástica de área total de circuito. Após modelagem e simulação para validação do bom funcionamento, a arquitetura proposta foi sintetizada utilizando recursos reconfiguráveis, do tipo FPGA. Os resultados são promissores. Abstract in english There are several neural network implementations using either software, hardware-based or a hardware/software co-design. This work proposes a hardware architecture to implement an artificial neural network (ANN), whose topology is the multilayer perceptron (MLP). In this paper, we explore the parall [...] elism of neural networks and allow on-thefly changes of the number of inputs, number of layers and number of neurons per layer of the net. This reconfigurability characteristic permits that any application of ANNs may be implemented using the proposed hardware. In order to reduce the processing time that is spent in arithmetic computation, a real number is represented using a fraction of integers. In this way, the arithmetics is limited to integer operations, performed by fast combinational circuits. A simple state machine is required to control sums and products of fractions. Sigmoid is used as the activation function in the proposed implementation. It is approximated by polynomials, whose underlying computation requires only sums and products. A theorem is introduced and proven so as to cover the arithmetic strategy of the computation of the activation function. Thus, the arithmetic circuitry used to implement the neuron weighted sum is reused for computing the sigmoid. this resource sharing decreased drastically the total area of the system. After modeling and simulation for functionality validation, the proposed architecture synthesized using reconfigurable hardware. The results are promising.

Rodrigo Martins da, Silva; Luiza de Macedo, Mourelle; Nadia, Nedjah.

33

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: English Abstract in portuguese Em termos computacionais, uma rede neural artificial (RNA) pode ser implementada em software ou em hardware, ou ainda de maneira híbrida, combinando ambos os recursos. O presente trabalho propõe uma arquitetura de hardware para a computação de uma rede neural do tipo perceptron com múltiplas camadas [...] (MLP). Soluções em hardware tendem a ser mais eficientes do que soluções em software. O projeto em questão, além de explorar fortemente o paralelismo das redes neurais, permite alterações do número de entradas, número de camadas e de neurônios por camada, de modo que diversas aplicações de RNAs possam ser executadas no hardware proposto. Visando a uma redução de tempo do processamento aritmético, um número real é aproximado por uma fração de inteiros. Dessa forma, as operações aritméticas limitam-se a operações inteiras, executadas por circuitos combinacionais. Uma simples máquina de estados é demandada para controlar somas e produtos de frações. A função de ativação usada neste projeto é a sigmóide. Essa função é aproximada mediante o uso de polinômios, cujas operações são regidas por somas e produtos. Um teorema é introduzido e provado, permitindo a fundamentação da estratégia de cálculo da função de ativação. Dessa forma, reaproveita-se o circuito aritmético da soma ponderada para também computar a sigmóide. Essa re-utilização dos recursos levou a uma redução drástica de área total de circuito. Após modelagem e simulação para validação do bom funcionamento, a arquitetura proposta foi sintetizada utilizando recursos reconfiguráveis, do tipo FPGA. Os resultados são promissores. Abstract in english There are several neural network implementations using either software, hardware-based or a hardware/software co-design. This work proposes a hardware architecture to implement an artificial neural network (ANN), whose topology is the multilayer perceptron (MLP). In this paper, we explore the parall [...] elism of neural networks and allow on-thefly changes of the number of inputs, number of layers and number of neurons per layer of the net. This reconfigurability characteristic permits that any application of ANNs may be implemented using the proposed hardware. In order to reduce the processing time that is spent in arithmetic computation, a real number is represented using a fraction of integers. In this way, the arithmetics is limited to integer operations, performed by fast combinational circuits. A simple state machine is required to control sums and products of fractions. Sigmoid is used as the activation function in the proposed implementation. It is approximated by polynomials, whose underlying computation requires only sums and products. A theorem is introduced and proven so as to cover the arithmetic strategy of the computation of the activation function. Thus, the arithmetic circuitry used to implement the neuron weighted sum is reused for computing the sigmoid. this resource sharing decreased drastically the total area of the system. After modeling and simulation for functionality validation, the proposed architecture synthesized using reconfigurable hardware. The results are promising.

Rodrigo Martins da, Silva; Luiza de Macedo, Mourelle; Nadia, Nedjah.

2011-12-01

34

Online learning dynamics of multilayer perceptrons with unidentifiable parameters

International Nuclear Information System (INIS)

In the over-realizable learning scenario of multilayer perceptrons, in which the student network has a larger number of hidden units than the true or optimal network, some of the weight parameters are unidentifiable. In this case, the teacher network consists of a union of optimal subspaces included in the parameter space. The optimal subspaces, which lead to singularities, are known to affect the estimation performance of neural networks. Using statistical mechanics, we investigate the online learning dynamics of two-layer neural networks in the over-realizable scenario with unidentifiable parameters. We show that the convergence speed strongly depends on the initial parameter conditions. We also show that there is a quasi-plateau around the optimal subspace, which differs from the well-known plateaus caused by permutation symmetry. In addition, we discuss the property of the final learning state, relating this to the singular structures

35

Training Algorithm for Extra Reduced Size Lattice-Ladder Multilayer Perceptrons

Digital Repository Infrastructure Vision for European Research (DRIVER)

A quick gradient training algorithm for a specific neural network structure called an extra reduced size lattice-ladder multilayer perceptron is introduced. Presented derivation of the algorithm utilizes recently found by author simplest way of exact computation of gradients for rotation parameters of lattice-ladder filter. Developed neural network training algorithm is optimal in terms of minimal number of constants, multiplication and addition operations, while the regularity of the structu...

Navakauskas, Dalius

2003-01-01

36

Multilayer perceptron in damage detection of bridge structures

Recent developments in artificial neural networks (ANN) have opened up new possibilities in the domain of structural engineering. For inverse problems like structural identification of large civil engineerlng structures such as bridges and buildings where the in situ measured data are expected to be imprecise and often incomplete, the ANN holds greater promise. The detection of structural damage and identification of damaged element in a large complex structure is a challenging task indeed. This paper presents an application of multilayer perceptron in the damage detection of steel bridge structures. The ssues relating to the design of network and learning paradigm are addressed and network architectures have been developed with reference to trussed bridge structures. The training patterns are generated for multiple damaged zones in a structure and performance of the networks with one and two hidden layers are examined. It has been observed that the performance of the network with two hidden layers was better than that of a single-layer architecture in general. The engineering importance of the whole exercise is demonstrated from the fact that measured input at only a few locations in the structure is needed in the identification process using the ANN.

Pandey, P. C.; Barai, S. V.

1995-02-01

37

Speeding up the Training of Lattice-Ladder Multilayer Perceptrons

Digital Repository Infrastructure Vision for European Research (DRIVER)

A lattice-ladder multilayer perceptron (LLMLP) is an appealing structure for advanced signal processing in a sense that it is nonlinear, possesses infinite impulse response and stability monitoring of it during training is simple. However, even moderate implementation of LLMLP training hinders the fact that a lot of storage and computation power must be allocated. In this paper we deal with the problem of computational efficiency of LLMLP training algorithms that are based on computation of g...

Navakauskas, Dalius

2002-01-01

38

Data classification with multilayer perceptrons using a generalized error function.

The learning process of a multilayer perceptron requires the optimization of an error function E(y,t) comparing the predicted output, y, and the observed target, t. We review some usual error functions, analyze their mathematical properties for data classification purposes, and introduce a new one, E(Exp), inspired by the Z-EDM algorithm that we have recently proposed. An important property of E(Exp) is its ability to emulate the behavior of other error functions by the sole adjustment of a real-valued parameter. In other words, E(Exp) is a sort of generalized error function embodying complementary features of other functions. The experimental results show that the flexibility of the new, generalized, error function allows one to obtain the best results achievable with the other functions with a performance improvement in some cases. PMID:18572384

Silva, Luís M; Marques de Sá, J; Alexandre, Luís A

2008-11-01

39

Fast parallel off-line training of multilayer perceptrons.

Various approaches to the parallel implementation of second-order gradient-based multilayer perceptron training algorithms are described. Two main classes of algorithm are defined involving Hessian and conjugate gradient-based methods. The limited- and full-memory Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithms are selected as representative examples and used to show that the step size and gradient calculations are critical components. For larger problems the matrix calculations in the full-memory algorithm are also significant. Various strategies are considered for parallelization, the best of which is implemented on parallel virtual machine (PVM) and transputer-based architectures. Results from a range of problems are used to demonstrate the performance achievable with each architecture. The transputer implementation is found to give excellent speed-ups but the problem size is limited by memory constraints. The speed-ups achievable with the PVM implementation are much poorer because of inefficient communication, but memory is not a difficulty. PMID:18255667

McLoone, S; Irwin, G W

1997-01-01

40

Second-Order Learning Methods for a Multilayer Perceptron

International Nuclear Information System (INIS)

First- and second-order learning methods for feed-forward multilayer neural networks are studied. Newton-type and quasi-Newton algorithms are considered and compared with commonly used back-propagation algorithm. It is shown that, although second-order algorithms require enhanced computer facilities, they provide better convergence and simplicity in usage. 13 refs., 2 figs., 2 tabs

41

International Nuclear Information System (INIS)

A comparative study of powers of multidimensional classifiers on the basis of integral nonparametric goodness-off-fit criteria ?nk and multilayer perceptrons was carried out in the cases where investigated distributions present simultaneous measurements of the same physical values in some detectors of experimental set-up. With the help of a numerical experiment it has been shown that multilayer perceptron provides the power close to limit if the identification of events is carried out in the space of effective variables. A procedure of transformation to such variables is described. Recommendations for the joint usage of the ?nk criteria and multilayer perceptrons are given. (author). 29 refs.; 15 figs

42

Missing value imputation on missing completely at random data using multilayer perceptrons.

Data mining is based on data files which usually contain errors in the form of missing values. This paper focuses on a methodological framework for the development of an automated data imputation model based on artificial neural networks. Fifteen real and simulated data sets are exposed to a perturbation experiment, based on the random generation of missing values. These data set sizes range from 47 to 1389 records. A perturbation experiment was performed for each data set where the probability of missing value was set to 0.05. Several architectures and learning algorithms for the multilayer perceptron are tested and compared with three classic imputation procedures: mean/mode imputation, regression and hot-deck. The obtained results, considering different performance measures, not only suggest this approach improves the quality of a database with missing values, but also the best results are clearly obtained using the Multilayer Perceptron model in data sets with categorical variables. Three learning rules (Levenberg-Marquardt, BFGS Quasi-Newton and Conjugate Gradient Fletcher-Reeves Update) and a small number of hidden nodes are recommended. PMID:20875726

Silva-Ramírez, Esther-Lydia; Pino-Mejías, Rafael; López-Coello, Manuel; Cubiles-de-la-Vega, María-Dolores

2011-01-01

43

FORECASTING ON FOREX MARKET WITH RBF AND PERCEPTRON NEURAL NETWORKS

Directory of Open Access Journals (Sweden)

Full Text Available This work deals with an alternative approach in financial modelling -artificial neural networks approach. The aim of this paper is to show that this type oftime series modelling is an excellent alternative to classical econometric modelling. Atfirst, neural networks using methods of supervised machine learning are discussed.After explaining theoretical basis of ANN, these models are then applied to specificexchange rate (AUD/USD. Finally, the comparison between statistical models andRBF and perceptron neural networks is made to illustrate the sense of using ANNmodels

ALEXANDRA KOTTILOVÁ

2012-01-01

44

In recent decades, the world has experienced unprecedented urban growth which endangers the green environment in and around urban areas. In this work, an artificial neural network (ANN) based model is developed to predict future impacts of urban and agricultural expansion on the uplands of Deepor Beel, a Ramsar wetland in the city area of Guwahati, Assam, India, by 2025 and 2035 respectively. Simulations were carried out for three different transition rates as determined from the changes during 2001-2011, namely simple extrapolation, Markov Chain (MC), and system dynamic (SD) modelling, using projected population growth, which were further investigated based on three different zoning policies. The first zoning policy employed no restriction while the second conversion restriction zoning policy restricted urban-agricultural expansion in the Guwahati Municipal Development Authority (GMDA) proposed green belt, extending to a third zoning policy providing wetland restoration in the proposed green belt. The prediction maps were found to be greatly influenced by the transition rates and the allowed transitions from one class to another within each sub-model. The model outputs were compared with GMDA land demand as proposed for 2025 whereby the land demand as produced by MC was found to best match the projected demand. Regarding the conservation of Deepor Beel, the Landscape Development Intensity (LDI) Index revealed that wetland restoration zoning policies may reduce the impact of urban growth on a local scale, but none of the zoning policies was found to minimize the impact on a broader base. The results from this study may assist the planning and reviewing of land use allocation within Guwahati city to secure ecological sustainability of the wetlands.

Mozumder, Chitrini; Tripathi, Nitin K.

2014-10-01

45

Ground Radar Target Classification Using Singular Value Decomposition and Multilayer Perceptron

Directory of Open Access Journals (Sweden)

Full Text Available The paper deals with classification of ground radar targets. Areceived radar signal backscattered from a ground radar target wasdigitized and in the form of radar signal matrix utilized for a featureextraction based on Singular Value Decomposition. Furthermore, singularvalues of a backscattered radar signal matrix, as a target feature,were utilized for Radar Target Classification by multilayer perceptron.In the learning phase of a multilayer perceptron we used the learningtarget set and in the testing phase the testing target set was used.The learning and testing target sets were created on the basis of realground radar targets.

I. Mokris

2001-12-01

46

Offensive and defensive systems of play represent important aspects of team sports. They include the players' positions at certain situations during a match, i.e., when players have to be on specific positions on the court. Patterns of play emerge based on the formations of the players on the court. Recognition of these patterns is important to react adequately and to adjust own strategies to the opponent. Furthermore, the ability to apply variable patterns of play seems to be promising since they make it harder for the opponent to adjust. The purpose of this study is to identify different team tactical patterns in volleyball and to analyze differences in variability. Overall 120 standard situations of six national teams in women's volleyball are analyzed during a world championship tournament. Twenty situations from each national team are chosen, including the base defence position (start configuration) and the two players block with middle back deep (end configuration). The shapes of the defence formations at the start and end configurations during the defence of each national team as well as the variability of these defence formations are statistically analyzed. Furthermore these shapes data are used to train multilayer perceptrons in order to test whether artificial neural networks can recognize the teams by their tactical patterns. Results show significant differences between the national teams in both the base defence position at the start and the two players block with middle back deep at the end of the standard defence situation. Furthermore, the national teams show significant differences in variability of the defence systems and start-positions are more variable than the end-positions. Multilayer perceptrons are able to recognize the teams at an average of 98.5%. It is concluded that defence systems in team sports are highly individual at a competitive level and variable even in standard situations. Artificial neural networks can be used to recognize teams by the shapes of the players' configurations. These findings support the concept that tactics and strategy have to be adapted for the team and need to be flexible in order to be successful. PMID:21414679

Jäger, Jörg M; Schöllhorn, Wolfgang I

2012-04-01

47

Estimate of significant wave height from non-coherent marine radar images by multilayer perceptrons

One of the most relevant parameters to characterize the severity of ocean waves is the significant wave height ( H s ). The estimate of H s from remotely sensed data acquired by non-coherent X-band marine radars is a problem not completely solved nowadays. A method commonly used in the literature (standard method) uses the square root of the signal-to-noise ratio (SNR) to linearly estimate H s . This method has been widely used during the last decade, but it presents some limitations, especially when swell-dominated sea states are present. To overcome these limitations, a new non-linear method incorporating additional sea state information is proposed in this article. This method is based on artificial neural networks (ANNs), specifically on multilayer perceptrons (MLPs). The information incorporated in the proposed MLP-based method is given by the wave monitoring system (WaMoS II) and concerns not only to the square root of the SNR, as in the standard method, but also to the peak wave length and mean wave period. Results for two different platforms (Ekofisk and FINO 1) placed in different locations of the North Sea are presented to analyze whether the proposed method works regardless of the sea states observed in each location or not. The obtained results empirically demonstrate how the proposed non-linear solution outperforms the standard method regardless of the environmental conditions (platform), maintaining real-time properties.

Vicen-Bueno, Raúl; Lido-Muela, Cristina; Nieto-Borge, José Carlos

2012-12-01

48

Belief Propagation for Error Correcting Codes and Lossy Compression Using Multilayer Perceptrons

Digital Repository Infrastructure Vision for European Research (DRIVER)

The belief propagation (BP) based algorithm is investigated as a potential decoder for both of error correcting codes and lossy compression, which are based on non-monotonic tree-like multilayer perceptron encoders. We discuss that whether the BP can give practical algorithms or not in these schemes. The BP implementations in those kind of fully connected networks unfortunately shows strong limitation, while the theoretical results seems a bit promising. Instead, it reveals ...

Mimura, Kazushi; Cousseau, Florent; Okada, Masato

2011-01-01

49

Classification of Parking Spots Using Multilayer Perceptron Networks

Directory of Open Access Journals (Sweden)

Full Text Available This project intends to develop a prototype for the identification of free spots in open air parking area where there is a good aerial view without obstacles, allowing for the identification of occupied and free spots. We used image processing techniques and pattern recognition using Artificial Neural Networks (ANN. In order to help implement the prototype, we used Matlab. In order to simulate the parking area, we created a model so that we could acquire the images using a webcam, process them, train the neural network, classify the spots and finally, show the results. The results show that it is viable to apply pattern recognition through image capture to classify parking spots

FALCAO, H. S.

2013-12-01

50

Classification of Parking Spots Using Multilayer Perceptron Networks

Digital Repository Infrastructure Vision for European Research (DRIVER)

This project intends to develop a prototype for the identification of free spots in open air parking area where there is a good aerial view without obstacles, allowing for the identification of occupied and free spots. We used image processing techniques and pattern recognition using Artificial Neural Networks (ANN). In order to help implement the prototype, we used Matlab. In order to simulate the parking area, we created a model so that we could acquire the images using a webcam, process th...

Falcao, H. S.; Lovato, A. V.; Dos, Santos A. F.; Oliveira, L. S.

2013-01-01

51

FPGA Implementation of Multilayer Perceptron for Modeling of Photovoltaic panel

International Nuclear Information System (INIS)

The Number of electronic applications using artificial neural network-based solutions has increased considerably in the last few years. However, their applications in photovoltaic systems are very limited. This paper introduces the preliminary result of the modeling and simulation of photovoltaic panel based on neural network and VHDL-language. In fact, an experimental database of meteorological data (irradiation, temperature) and output electrical generation signals of the PV-panel (current and voltage) has been used in this study. The inputs of the ANN-PV-panel are the daily total irradiation and mean average temperature while the outputs are the current and voltage generated from the panel. Firstly, a dataset of 4x364 have been used for training the network. Subsequently, the neural network (MLP) corresponding to PV-panel is simulated using VHDL language based on the saved weights and bias of the network. Simulation results of the trained MLP-PV panel based on Matlab and VHDL are presented. The proposed PV-panel model based ANN and VHDL permit to evaluate the performance PV-panel using only the environmental factors and involves less computational efforts, and it can be used for predicting the output electrical energy from the PV-panel

52

Fuzzy and Multilayer Perceptron for Evaluation of HV Bushings

The work proposes the application of fuzzy set theory (FST) to diagnose the condition of high voltage bushings. The diagnosis uses dissolved gas analysis (DGA) data from bushings based on IEC60599 and IEEE C57-104 criteria for oil impregnated paper (OIP) bushings. FST and neural networks are compared in terms of accuracy and computational efficiency. Both FST and NN simulations were able to diagnose the bushings condition with 10% error. By using fuzzy theory, the maintenance department can classify bushings and know the extent of degradation in the component.

Dhlamini, Sizwe M; Majozi, Thokozani

2007-01-01

53

Belief Propagation for Error Correcting Codes and Lossy Compression Using Multilayer Perceptrons

The belief propagation (BP) based algorithm is investigated as a potential decoder for both of error correcting codes and lossy compression, which are based on non-monotonic tree-like multilayer perceptron encoders. We discuss that whether the BP can give practical algorithms or not in these schemes. The BP implementations in those kind of fully connected networks unfortunately shows strong limitation, while the theoretical results seems a bit promising. Instead, it reveals it might have a rich and complex structure of the solution space via the BP-based algorithms.

Mimura, Kazushi; Okada, Masato

2011-01-01

54

Belief Propagation for Error Correcting Codes and Lossy Compression Using Multilayer Perceptrons

The belief propagation (BP) based algorithm is investigated as a potential decoder for both of error correcting codes and lossy compression, which are based on non-monotonic tree-like multilayer perceptron encoders. We discuss that whether the BP can give practical algorithms or not in these schemes. The BP implementations in those kind of fully connected networks unfortunately shows strong limitation, while the theoretical results seems a bit promising. Instead, it reveals it might have a rich and complex structure of the solution space via the BP-based algorithms.

Mimura, Kazushi; Cousseau, Florent; Okada, Masato

2011-03-01

55

Directory of Open Access Journals (Sweden)

Full Text Available In order to perceive of rainfall- runoff process, essential prediction for water surface source management has special importance. Thereby in this paper, Tang-e Karzin hydrometric station which is located in sub-domain of salman-farsi dam had been considered. By utilizing of weekly statistical discharge information related to past 36 years, multilayer perceptron neural network model was used to predict the average weekly discharge of Tang-e Karzin station through the discharge information of its two upside stations. In order to optimize the weights and biases of the MLP network, we tried to use Artificial Bee Colony (ABC algorithm within training phase of the ANN network. The results indicated that by changing of different parameters of hidden layer of perceptron model, ABC can well optimize ANN’s weights and biases. Among five activation function Log-sigmoid was performed better than others with 9 neurons in hidden layer

Saleh Salimi

2013-10-01

56

Image Binarization Using Multi-Layer Perceptron: A Semi-Supervised Approach

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, we have discussed the Image Binarization technique using Multilayer Perceptron (MLP. The purpose of Image Binarization is to extract the lightness (brightness, density as a feature amount from the Image. It converts a gray-scale image of up to 256 gray levels to a black and white image. We use Backpropagation algorithm for training MLP. It is a supervised learning technique. Here Kmeans clustering algorithm has been used for clustering a 256 × 256 gray-level image. The dataset obtained by this is fed to the MLP and processed in a Semi-Supervised way where some training samples are taken as Known patterns (for training and others as Unknown patterns. Finally through this approach a Binarized image is produced.

Amlan Raychaudhuri

2012-04-01

57

Static Digits Recognition Using Rotational Signatures and Hu Moments with a Multilayer Perceptron

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents two systems for recognizing static signs (digits from American Sign Language (ASL. These systems avoid the use color marks, or gloves, using instead, low-pass and high-pass filters in space and frequency domains, and color space transformations. First system used rotational signatures based on a correlation operator; minimum distance was used for the classification task. Second system computed the seven Hu invariants from binary images; these descriptors fed to a Multi-Layer Perceptron (MLP in order to recognize the 9 different classes. First system achieves 100% of recognition rate with leaving-one-out validation and second experiment performs 96.7% of recognition rate with Hu moments and 100% using 36 normalized moments and k-fold cross validation.

Francisco Solís

2014-10-01

58

Time series modeling with pruned multi-layer perceptron and 2-stage damped least-squares method

A Multi-Layer Perceptron (MLP) defines a family of artificial neural networks often used in TS modeling and forecasting. Because of its "black box" aspect, many researchers refuse to use it. Moreover, the optimization (often based on the exhaustive approach where "all" configurations are tested) and learning phases of this artificial intelligence tool (often based on the Levenberg-Marquardt algorithm; LMA) are weaknesses of this approach (exhaustively and local minima). These two tasks must be repeated depending on the knowledge of each new problem studied, making the process, long, laborious and not systematically robust. In this paper a pruning process is proposed. This method allows, during the training phase, to carry out an inputs selecting method activating (or not) inter-nodes connections in order to verify if forecasting is improved. We propose to use iteratively the popular damped least-squares method to activate inputs and neurons. A first pass is applied to 10% of the learning sample to determine weights significantly different from 0 and delete other. Then a classical batch process based on LMA is used with the new MLP. The validation is done using 25 measured meteorological TS and cross-comparing the prediction results of the classical LMA and the 2-stage LMA.

Voyant, Cyril; Tamas, Wani; Paoli, Christophe; Balu, Aurélia; Muselli, Marc; Nivet, Marie-Laure; Notton, Gilles

2014-03-01

59

Digital Repository Infrastructure Vision for European Research (DRIVER)

This paper presents a novel land cover change detection method that employs a sliding window over hyper-temporal multi-spectral images acquired from the 7 bands of the MODerate-resolution Imaging Spectroradiometer (MODIS) land surface reflectance product. The method uses a Feedforward Multilayer Perceptron (MLP) for supervised change detection that operates on multi-spectral time series extracted with a sliding window from the dataset. The method was evaluated on both real and simulated land ...

Salmon, Brian Paxton; Olivier, Jan Corne; Kleynhans, Waldo; Wessels, Konrad J.; Den Bergh, Frans; Steenkamp, Karen C.

2011-01-01

60

Directory of Open Access Journals (Sweden)

Full Text Available Seismic object detection is a relatively new field in which 3-D bodies are visualized and spatial relationships between objects of different origins are studied in order to extract geologic information. In this paper, we propose a method for finding an optimal classifier with the help of a statistical feature ranking technique and combining different classifiers. The method, which has general applicability, is demonstrated here on a gas chimney detection problem. First, we evaluate a set of input seismic attributes extracted at locations labeled by a human expert using regularized discriminant analysis (RDA. In order to find the RDA score for each seismic attribute, forward and backward search strategies are used. Subsequently, two non-linear classifiers: multilayer perceptron (MLP and support vector classifier (SVC are run on the ranked seismic attributes. Finally, to capitalize on the intrinsic differences between both classifiers, the MLP and SVC results are combined using logical rules of maximum, minimum and mean. The proposed method optimizes the ranked feature space size and yields the lowest classification error in the final combined result. We will show that the logical minimum reveals gas chimneys that exhibit both the softness of MLP and the resolution of SVC classifiers.

H. Hashemi

2008-11-01

61

Seismic object detection is a relatively new field in which 3-D bodies are visualized and spatial relationships between objects of different origins are studied in order to extract geologic information. In this paper, we propose a method for finding an optimal classifier with the help of a statistical feature ranking technique and combining different classifiers. The method, which has general applicability, is demonstrated here on a gas chimney detection problem. First, we evaluate a set of input seismic attributes extracted at locations labeled by a human expert using regularized discriminant analysis (RDA). In order to find the RDA score for each seismic attribute, forward and backward search strategies are used. Subsequently, two non-linear classifiers: multilayer perceptron (MLP) and support vector classifier (SVC) are run on the ranked seismic attributes. Finally, to capitalize on the intrinsic differences between both classifiers, the MLP and SVC results are combined using logical rules of maximum, minimum and mean. The proposed method optimizes the ranked feature space size and yields the lowest classification error in the final combined result. We will show that the logical minimum reveals gas chimneys that exhibit both the softness of MLP and the resolution of SVC classifiers.

Hashemi, H.; Tax, D. M. J.; Duin, R. P. W.; Javaherian, A.; de Groot, P.

2008-11-01

62

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents the results obtained by developing a methodology to detect 5 types of heartbeats (Normal (N, Right bundle branch block (RBBB, Left bundle branch block (LBBB, Premature atrial contraction (APC and Premature ventricular contraction (PVC, using Wavelet transform packets with non-adaptative mode applied on features extraction from heartbeats. It was used the Shannon function to calculate the entropy and It was added an identification nodes stage per every type of cardiac signal in the Wavelet tree. The using of Wavelet packets transform allows the access to information which results of decomposition of low and high frecuency, giving providing a more integral analysis than achieved by the discrete Wavelet transform. Three families of mother Wavelet were evaluated on transformation: Daubechies, Symlet and Reverse Biorthogonal, which were results from a previous research in that were identified the mother Wavelet that had higher entropy with the cardiac signals. With non-adaptive mode, the computational cost is reduced when Wavelet packets are used; this cost represents the most marked disadvantage from the transform. To classify the heartbeats were used Support Vector Machines and Multilayer Perceptron. The best classification error was achieved employing Support Vector Machine and a radial basis function; it was 2.57 %.

Alejandro J. Orozco-Naranjo

2013-11-01

63

In this paper, we propose a wideband dynamic behavioral model for a bulk reflective semiconductor optical amplifier (RSOA) used as a modulator in colorless radio over fiber (RoF) systems using a tapped-delay multilayer perceptron (TDMLP). 64 quadrature amplitude modulation (QAM) signals with 20 Msymbol/s were used to train, validate and test the model. Nonlinear distortion and dynamic effects induced by the RSOA modulator are demonstrated. The parameters of the model such as the number of nodes in the hidden layer and memory depth were optimized to ensure the generality and accuracy. The normalized mean square error (NMSE) is used as a figure of merit. The NMSE was up to -44.33 dB when the number of nodes in the hidden layer and memory depth were set to 20 and 3, respectively. The TDMLP model can accurately approximate to the dynamic characteristics of the RSOA modulator. The dynamic AM-AM and dynamic AM-PM distortions of the RSOA modulator are drawn. The results show that the single hidden layer TDMLP can provide accurate approximation for behaviors of the RSOA modulator. PMID:23481795

Liu, Zhansheng; Violas, Manuel Alberto; Carvalho, Nuno Borges

2013-02-11

64

Directory of Open Access Journals (Sweden)

Full Text Available Normalization is important for Electrical Capacitance Tomography (ECT data due to the very small capacitance values obtained either from the physical or simulated ECT system. Thus far, there are two commonly used normalization methods for ECT, but their suitability has not been investigated. This paper presents the work on comparing the performances of two Multilayer Perceptron (MLP neural networks; one trained based on ECT data normalized using the conventional equation and the other normalized using the improved equation, to recognize gas-oil flow patterns. The correct pattern recognition percentages for both MLPs were calculated and compared. The results showed that the MLP trained with the conventional ECT normalization equation out-performed the ones trained with the improved normalization data for the task of gas-oil pattern recognition.

Hafizah Talib

2009-02-01

65

Storage capacity of correlated perceptrons

We consider an ensemble of $K$ single-layer perceptrons exposed to random inputs and investigate the conditions under which the couplings of these perceptrons can be chosen such that prescribed correlations between the outputs occur. A general formalism is introduced using a multi-perceptron costfunction that allows to determine the maximal number of random inputs as a function of the desired values of the correlations. Replica-symmetric results for $K=2$ and $K=3$ are compared with properties of two-layer networks of tree-structure and fixed Boolean function between hidden units and output. The results show which correlations in the hidden layer of multi-layer neural networks are crucial for the value of the storage capacity.

Malzahn, D; Kanter, Yu

1996-01-01

66

Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise

Iterative gradient methods such as Levenberg-Marquardt (LM) are in widespread use for source localization from electroencephalographic (EEG) and magnetoencephalographic (MEG) signals. Unfortunately, LM depends sensitively on the initial guess, necessitating repeated runs. This, combined with LM's high per-step cost, makes its computational burden quite high. To reduce this burden, we trained a multilayer perceptron (MLP) as a real-time localizer. We used an analytical model of quasistatic electromagnetic propagation through a spherical head to map randomly chosen dipoles to sensor activities according to the sensor geometry of a 4D Neuroimaging Neuromag-122 MEG system, and trained a MLP to invert this mapping in the absence of noise or in the presence of various sorts of noise such as white Gaussian noise, correlated noise, or real brain noise. A MLP structure was chosen to trade off computation and accuracy. This MLP was trained four times, with each type of noise. We measured the effects of initial guesses on LM performance, which motivated a hybrid MLP-start-LM method, in which the trained MLP initializes LM. We also compared the localization performance of LM, MLPs, and hybrid MLP-start-LMs for realistic brain signals. Trained MLPs are much faster than other methods, while the hybrid MLP-start-LMs are faster and more accurate than fixed-4-start-LM. In particular, the hybrid MLP-start-LM initialized by a MLP trained with the real brain noise dataset is 60 times faster and is comparable in accuracy to random-20-start-LM, and this hybrid system (localization error: 0.28 cm, computation time: 36 ms) shows almost as good performance as optimal-1-start-LM (localization error: 0.23 cm, computation time: 22 ms), which initializes LM with the correct dipole location. MLPs trained with noise perform better than the MLP trained without noise, and the MLP trained with real brain noise is almost as good an initial guesser for LM as the correct dipole location.

Jun, Sung Chan; Pearlmutter, Barak A.; Nolte, Guido

2002-07-01

67

Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise

International Nuclear Information System (INIS)

Iterative gradient methods such as Levenberg-Marquardt (LM) are in widespread use for source localization from electroencephalographic (EEG) and magnetoencephalographic (MEG) signals. Unfortunately, LM depends sensitively on the initial guess, necessitating repeated runs. This, combined with LM's high per-step cost, makes its computational burden quite high. To reduce this burden, we trained a multilayer perceptron (MLP) as a real-time localizer. We used an analytical model of quasistatic electromagnetic propagation through a spherical head to map randomly chosen dipoles to sensor activities according to the sensor geometry of a 4D Neuroimaging Neuromag-122 MEG system, and trained a MLP to invert this mapping in the absence of noise or in the presence of various sorts of noise such as white Gaussian noise, correlated noise, or real brain noise. A MLP structure was chosen to trade off computation and accuracy. This MLP was trained four times, with each type of noise. We measured the effects of initial guesses on LM performance, which motivated a hybrid MLP-start-LM method, in which the trained MLP initializes LM. We also compared the localization performance of LM, MLPs, and hybrid MLP-start-LMs for realistic brain signals. Trained MLPs are much faster than other methods, while the hybrid MLP-start-LMs are faster and more accurate than fixed-4-start-LM. In particular, the hybrid MLP-start-LM initialized by a MLP trained with the real brain noise dataset is 60 times faster and is comparable in accuracy to random-20-start-LM, and this hybrid system (localization error: 0.28 cm, computation time: 36 ms) shows almost as good performance as optimal-1-start-LM (localization error: 0.23 cm, computation time: 22 ms), which initializes LM with the correct dipole location. MLPs trained with noise perform better than the MLP trained without noise, and the MLP trained with real brain noise is almost as good an initial guesser for LM as the correct dipole location. (author) )

68

Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise

Energy Technology Data Exchange (ETDEWEB)

Iterative gradient methods such as Levenberg-Marquardt (LM) are in widespread use for source localization from electroencephalographic (EEG) and magnetoencephalographic (MEG) signals. Unfortunately, LM depends sensitively on the initial guess, necessitating repeated runs. This, combined with LM's high per-step cost, makes its computational burden quite high. To reduce this burden, we trained a multilayer perceptron (MLP) as a real-time localizer. We used an analytical model of quasistatic electromagnetic propagation through a spherical head to map randomly chosen dipoles to sensor activities according to the sensor geometry of a 4D Neuroimaging Neuromag-122 MEG system, and trained a MLP to invert this mapping in the absence of noise or in the presence of various sorts of noise such as white Gaussian noise, correlated noise, or real brain noise. A MLP structure was chosen to trade off computation and accuracy. This MLP was trained four times, with each type of noise. We measured the effects of initial guesses on LM performance, which motivated a hybrid MLP-start-LM method, in which the trained MLP initializes LM. We also compared the localization performance of LM, MLPs, and hybrid MLP-start-LMs for realistic brain signals. Trained MLPs are much faster than other methods, while the hybrid MLP-start-LMs are faster and more accurate than fixed-4-start-LM. In particular, the hybrid MLP-start-LM initialized by a MLP trained with the real brain noise dataset is 60 times faster and is comparable in accuracy to random-20-start-LM, and this hybrid system (localization error: 0.28 cm, computation time: 36 ms) shows almost as good performance as optimal-1-start-LM (localization error: 0.23 cm, computation time: 22 ms), which initializes LM with the correct dipole location. MLPs trained with noise perform better than the MLP trained without noise, and the MLP trained with real brain noise is almost as good an initial guesser for LM as the correct dipole location. (author)

Jun, Sung Chan [Department of Computer Science, University of New Mexico, Albuquerque, NM (Mexico)]. E-mail: junsc@cs.unm.edu; Pearlmutter, Barak A.; Nolte, Guido [Department of Computer Science, University of New Mexico, Albuquerque, NM (Mexico)

2002-07-21

69

Universal perceptron (UP), a generalization of Rosenblatt's perceptron, is considered in this paper, which is capable of implementing all Boolean functions (BFs). In the classification of BFs, there are: 1) linearly separable Boolean function (LSBF) class, 2) parity Boolean function (PBF) class, and 3) non-LSBF and non-PBF class. To implement these functions, UP takes different kinds of simple topological structures in which each contains at most one hidden layer along with the smallest possible number of hidden neurons. Inspired by the concept of DNA sequences in biological systems, a novel learning algorithm named DNA-like learning is developed, which is able to quickly train a network with any prescribed BF. The focus is on performing LSBF and PBF by a single-layer perceptron (SLP) with the new algorithm. Two criteria for LSBF and PBF are proposed, respectively, and a new measure for a BF, named nonlinearly separable degree (NLSD), is introduced. In the sense of this measure, the PBF is the most complex one. The new algorithm has many advantages including, in particular, fast running speed, good robustness, and no need of considering the convergence property. For example, the number of iterations and computations in implementing the basic 2-bit logic operations such as AND, OR, and XOR by using the new algorithm is far smaller than the ones needed by using other existing algorithms such as error-correction (EC) and backpropagation (BP) algorithms. Moreover, the synaptic weights and threshold values derived from UP can be directly used in designing of the template of cellular neural networks (CNNs), which has been considered as a new spatial-temporal sensory computing paradigm. PMID:23460987

Chen, Fangyue; Chen, Guanrong Ron; He, Guolong; Xu, Xiubin; He, Qinbin

2009-10-01

70

Optical scattering spectra obtained in the clinical trials of breast cancer diagnostic system were analyzed for the purpose to detect in the dataflow the segments corresponding to malignant tissues. Minimal invasive probe with optical fibers inside delivers white light from the source and collects the scattering light while being moved through the tissue. The sampling rate is 100 Hz and each record contains the results of measurements of scattered light intensity at 184 fixed wavelength points. Large amount of information acquired in each procedure, fuzziness in criteria of 'cancer' family membership and data noisiness make neural networks to be an attractive tool for analysis of these data. To define the dividing rule between 'cancer' and 'non-cancer' spectral families a three-layer perceptron was applied. In the process of perceptron learning back propagation method was used to minimize the learning error. Regularization was done using the Bayesian approach. The learning sample was formed by the experts. End-to-end probability calculation throughout the procedure dataset showed reliable detection of the 'cancer' segments. Much attention was paid on the spectra of the tissues with high blood content. Often the reason is vessel injury caused by the penetrating optical probe. But also it can be a dense vessel net surrounding the malignant tumor. To make the division into 'cancer' and 'non-cancer' families for the tissues with high blood content a special perceptron was learnt exceptionally on such spectra.

Nuzhny, Anton S.; Shumsky, Sergey A.; Korzhov, Alexey G.; Lyubynskaya, Tatiana E.

2008-02-01

71

Replica Symmetry Breaking and the Kuhn-Tucker Cavity Method in simple and multilayer Perceptrons

Within a Kuhn-Tucker cavity method introduced in a former paper, we study optimal stability learning for situations, where in the replica formalism the replica symmetry may be broken, namely (i) the case of a simple perceptron above the critical loading, and (ii) the case of two-layer AND-perceptrons, if one learns with maximal stability. We find that the deviation of our cavity solution from the replica symmetric one in these cases is a clear indication of the necessity of replica symmetry breaking. In any case the cavity solution tends to underestimate the storage capabilities of the networks.

Gerl, F

1996-01-01

72

On-line Gibbs learning. II. Application to perceptron and multilayer networks

In the preceding paper (``On-line Gibbs Learning. I. General Theory'') we have presented the on-line Gibbs algorithm (OLGA) and studied analytically its asymptotic convergence. In this paper we apply OLGA to on-line supervised learning in several network architectures: a single-layer perceptron, two-layer committee machine, and a winner-takes-all (WTA) classifier. The behavior of OLGA for a single-layer perceptron is studied both analytically and numerically for a variety of rules: a realizable perceptron rule, a perceptron rule corrupted by output and input noise, and a rule generated by a committee machine. The two-layer committee machine is studied numerically for the cases of learning a realizable rule as well as a rule that is corrupted by output noise. The WTA network is studied numerically for the case of a realizable rule. The asymptotic results reported in this paper agree with the predictions of the general theory of OLGA presented in paper I. In all the studied cases, OLGA converges to a set of weights that minimizes the generalization error. When the learning rate is chosen as a power law with an optimal power, OLGA converges with a power law that is the same as that of batch learning.

Kim, J. W.; Sompolinsky, H.

1998-08-01

73

The random initialization of weights of a multilayer perceptron makes it possible to model its training process as a Las Vegas algorithm, i.e. a randomized algorithm which stops when some required training error is obtained, and whose execution time is a random variable. This modelling is used to perform a case study on a well-known pattern recognition benchmark: the UCI Thyroid Disease Database. Empirical evidence is presented of the training time probability distribution exhibiting a heavy tail behavior, meaning a big probability mass of long executions. This fact is exploited to reduce the training time cost by applying two simple restart strategies. The first assumes full knowledge of the distribution yielding a 40% cut down in expected time with respect to the training without restarts. The second, assumes null knowledge, yielding a reduction ranging from 9% to 23%.

Cebrian, Manuel

2007-01-01

74

Storage capacity of correlated perceptrons

We consider an ensemble of K single-layer perceptrons exposed to random inputs and investigate the conditions under which the couplings of these perceptrons can be chosen such that prescribed correlations between the outputs occur. A general formalism is introduced using a multiperceptron cost function that allows one to determine the maximal number of random inputs as a function of the desired values of the correlations. Replica-symmetric results for K=2 and K=3 are compared with properties of two-layer networks of tree-structure and fixed Boolean function between hidden units and output. The results show which correlations in the hidden layer of multilayer neural networks are crucial for the value of the storage capacity.

Malzahn, D.; Engel, A.; Kanter, I.

1997-06-01

75

Visualization of learning in multilayer perceptron networks using principal component analysis.

This paper is concerned with the use of scientific visualization methods for the analysis of feedforward neural networks (NNs). Inevitably, the kinds of data associated with the design and implementation of neural networks are of very high dimensionality, presenting a major challenge for visualization. A method is described using the well-known statistical technique of principal component analysis (PCA). This is found to be an effective and useful method of visualizing the learning trajectories of many learning algorithms such as backpropagation and can also be used to provide insight into the learning process and the nature of the error surface. PMID:18238154

Gallagher, M; Downs, T

2003-01-01

76

In this letter, we use the firing rates from an array of olfactory sensory neurons (OSNs) of the fruit fly, Drosophila melanogaster, to train an artificial neural network (ANN) to distinguish different chemical classes of volatile odorants. Bootstrapping is implemented for the optimized networks, providing an accurate estimate of a network's predicted values. Initially a simple linear predictor was used to assess the complexity of the data and was found to provide low prediction performance. A nonlinear ANN in the form of a single multilayer perceptron (MLP) was also used, providing a significant increase in prediction performance. The effect of the number of hidden layers and hidden neurons of the MLP was investigated and found to be effective in enhancing network performance with both a single and a double hidden layer investigated separately. A hybrid array of MLPs was investigated and compared against the single MLP architecture. The hybrid MLPs were found to classify all vectors of the validation set, presenting the highest degree of prediction accuracy. Adjustment of the number of hidden neurons was investigated, providing further performance gain. In addition, noise injection was investigated, proving successful for certain network designs. It was found that the best-performing MLP was that of the double-hidden-layer hybrid MLP network without the use of noise injection. Furthermore, the level of performance was examined when different numbers of OSNs used were varied from the maximum of 24 to only 5 OSNs. Finally, the ideal OSNs were identified that optimized network performance. The results obtained from this study provide strong evidence of the usefulness of ANNs in the field of olfaction for the future realization of a signal processing back end for an artificial olfactory biosensor. PMID:23020109

Bachtiar, Luqman R; Unsworth, Charles P; Newcomb, Richard D; Crampin, Edmund J

2013-01-01

77

Training multi-layered neural network neocognitron.

This paper proposes new learning rules suited for training multi-layered neural networks and applies them to the neocognitron. The neocognitron is a hierarchical multi-layered neural network capable of robust visual pattern recognition. It acquires the ability to recognize visual patterns through learning. For training intermediate layers of the hierarchical network of the neocognitron, we use a new learning rule named add-if-silent. By the use of the add-if-silent rule, the learning process becomes much simpler and more stable, and the computational cost for learning is largely reduced. Nevertheless, a high recognition rate can be kept without increasing the scale of the network. For the highest stage of the network, we use the method of interpolating-vector. We have previously reported that the recognition rate is greatly increased if this method is used during recognition. This paper proposes a new method of using it for both learning and recognition. Computer simulation demonstrates that the new neocognitron, which uses the add-if-silent and the interpolating-vector, produces a higher recognition rate for handwritten digits recognition with a smaller scale of the network than the neocognitron of previous versions. PMID:23380595

Fukushima, Kunihiko

2013-04-01

78

Membership generation using multilayer neural network

There has been intensive research in neural network applications to pattern recognition problems. Particularly, the back-propagation network has attracted many researchers because of its outstanding performance in pattern recognition applications. In this section, we describe a new method to generate membership functions from training data using a multilayer neural network. The basic idea behind the approach is as follows. The output values of a sigmoid activation function of a neuron bear remarkable resemblance to membership values. Therefore, we can regard the sigmoid activation values as the membership values in fuzzy set theory. Thus, in order to generate class membership values, we first train a suitable multilayer network using a training algorithm such as the back-propagation algorithm. After the training procedure converges, the resulting network can be treated as a membership generation network, where the inputs are feature values and the outputs are membership values in the different classes. This method allows fairly complex membership functions to be generated because the network is highly nonlinear in general. Also, it is to be noted that the membership functions are generated from a classification point of view. For pattern recognition applications, this is highly desirable, although the membership values may not be indicative of the degree of typicality of a feature value in a particular class.

Kim, Jaeseok

1992-01-01

79

A new and novel training algorithm, based upon the matrix pseudoinverse least-squares method, is introduced for training hidden layer, forward-feed neural networks with high accuracy and speed for nonlinear and chaotic time series prediction. Model-generated chaotic time series, including that of the Lorenz system, are used to measure performance and robustness. Our new training algorithm has rendered application of forward-feed, hidden-layer neural networks for adaptive chaotic time series analysis, as well as other signal processing, practical and near real time using standard desktop computation facilities. We have applied our method, in conjunction with other standard methods, to the analysis of stimulated Brillouin scattering under cw pump conditions involving a single Stokes and pump signal in a single- mode optical fiber as the nonlinear medium. We use Stokes signal data generated from a standard model and correlate the training performance of our algorithm with statistical and dynamical characteristics of the system determined by other means.

Pethel, Shawn D.; Bowden, Charles M.; Scalora, Michael

1993-12-01

80

Electron/pion identification in the CBM TRD using a multilayer perceptron

International Nuclear Information System (INIS)

The problem of electron/pion identification in the CBM experiment based on the measurements of energy losses and transition radiation in the TRD detector is discussed. A possibility to solve such a problem by applying an artificial neural network (ANN) is considered. As input information for the network we used both the samples of energy losses of pions or electrons in the TRD absorbers and the 'clever' variable obtained on the basis of the original data. We show that usage of this new variable permits one to reach a reliable level of particle recognition no longer than after 10-20 training epochs; there are practically no fluctuations against the trend, and the needed level of pions suppression is obtained under the condition of a minimal loss of electrons

81

A learning rule for very simple universal approximators consisting of a single layer of perceptrons.

One may argue that the simplest type of neural networks beyond a single perceptron is an array of several perceptrons in parallel. In spite of their simplicity, such circuits can compute any Boolean function if one views the majority of the binary perceptron outputs as the binary output of the parallel perceptron, and they are universal approximators for arbitrary continuous functions with values in [0,1] if one views the fraction of perceptrons that output 1 as the analog output of the parallel perceptron. Note that in contrast to the familiar model of a "multi-layer perceptron" the parallel perceptron that we consider here has just binary values as outputs of gates on the hidden layer. For a long time one has thought that there exists no competitive learning algorithm for these extremely simple neural networks, which also came to be known as committee machines. It is commonly assumed that one has to replace the hard threshold gates on the hidden layer by sigmoidal gates (or RBF-gates) and that one has to tune the weights on at least two successive layers in order to achieve satisfactory learning results for any class of neural networks that yield universal approximators. We show that this assumption is not true, by exhibiting a simple learning algorithm for parallel perceptrons - the parallel delta rule (p-delta rule). In contrast to backprop for multi-layer perceptrons, the p-delta rule only has to tune a single layer of weights, and it does not require the computation and communication of analog values with high precision. Reduced communication also distinguishes our new learning rule from other learning rules for parallel perceptrons such as MADALINE. Obviously these features make the p-delta rule attractive as a biologically more realistic alternative to backprop in biological neural circuits, but also for implementations in special purpose hardware. We show that the p-delta rule also implements gradient descent-with regard to a suitable error measure-although it does not require to compute derivatives. Furthermore it is shown through experiments on common real-world benchmark datasets that its performance is competitive with that of other learning approaches from neural networks and machine learning. It has recently been shown [Anthony, M. (2007). On the generalization error of fixed combinations of classifiers. Journal of Computer and System Sciences 73(5), 725-734; Anthony, M. (2004). On learning a function of perceptrons. In Proceedings of the 2004 IEEE international joint conference on neural networks (pp. 967-972): Vol. 2] that one can also prove quite satisfactory bounds for the generalization error of this new learning rule. PMID:18249524

Auer, Peter; Burgsteiner, Harald; Maass, Wolfgang

2008-06-01

82

Object Recognition Using Multi-Layer Hopfield Neural Network.

An object recognition approach based on concurrent coarse-and-fine matching using a multi-layer Hopfield neural network is presented. The proposed network consists of several cascaded single layer Hopfield networks, each encoding object features at a dist...

S. S. Young, P. D. Scott, N. M. Nasrabadi

1994-01-01

83

Directory of Open Access Journals (Sweden)

Full Text Available When investors decide to “adventure” through stock markets they search for a method to provide safety on making decision. In fact, there is no precise way to know which stocks will became a profitable investiment. Technical analysis is a discipline that support the investors on making decisions. Such a discipline uses a set of tools and statistical methods to forecast the market’s movement. Such a paper presents the develpment of a robotical Trade System, using a heuristic method. The system has a Neural Network multilayer perceptron, trained with an algorithm for back propagation error. Thus, approaching to the technical analysis without emotional aspects, using the Neural Network forecast on supporting the decisions of a investor on stock market. In analyzing the results of the neural network can be seen that the neural network got a result of 42.6% higher than the diagnostic of the technical analysis.Quando investidores decidem se “aventurar” pelo mercado de renda variável, como pelo mercado de ações, buscam um método de ter mais segurança na tomada de decisão. Na prática, não há como saber quais ativos tornar-se-ão um investimento lucrativo. No mercado acionário, a Análise Técnica procura auxiliar o investidor na tomada de decisão. Para isso, utiliza-se de ferramentas e de métodos estatísticos para tentar predizer os movimentos do mercado. Este artigo apresenta o desenvolvimento de um Trade System robótico, utilizando um método heurístico. O sistema conta com uma rede neural multilayer perceptron, treinada com o algoritmo de retro propagação de erro, aproximando-se da análise técnica sem o fator emoção. Ao avaliar os resultados da rede neural, pode ser visto que a mesma obteve um resultado de 42,6% maior do que o diagnóstico da análise técnica.

Vânia Medianeira Flores Costa

2012-04-01

84

In this paper, the weights of a Neural Network using Chaotic Imperialist Competitive Algorithm are updated. A three-layered Perseptron Neural Network applied for prediction of the maximum worth of the stocks changed in TEHRAN's bourse market. We trained this neural network with CICA, ICA, PSO and GA algorithms and compared the results with each other. The consideration of the results showed that the training and test error of the network trained by the CICA algorithm has been reduced in comparison to the other three methods.

Zhang, Xiuping

85

Scientific Electronic Library Online (English)

Full Text Available SciELO Cuba | Language: Spanish Abstract in spanish El perceptrón multicapa (PMC) figura dentro de los tipos de redes neuronales artificiales (RNA) con resultados útiles en los estudios de relación estructura-actividad. Dado que los volúmenes de datos en proyectos de Bioinformática son eventualmente grandes, se propuso evaluar algoritmos para acortar [...] el tiempo de entrenamiento de la red sin afectar su eficiencia. Se desarrolló un algoritmo para el entrenamiento local y distribuido del PMC con la posibilidad de variar las funciones de transferencias para lo cual se utilizaron el Weka y la Plataforma de Tareas Distribuidas Tarenal para distribuir el entrenamiento del perceptrón multicapa. Se demostró que en dependencia de la muestra de entrenamiento, la variación de las funciones de transferencia pueden reportar resultados mucho más eficientes que los obtenidos con la clásica función Sigmoidal, con incremento de la g-media entre el 4.5 y el 17 %. Se encontró además que en los entrenamientos distribuidos es posible alcanzar eventualmente mejores resultados que los logrados en ambiente local. Abstract in english The multilayer perceptron (PMC) ranks among the types of artificial neural networks (ANN), which has provided better results in studies of structure-activity relationship. As the data volumes in Bioinformatics' projects are eventually big, it was proposed to evaluate algorithms to shorten the traini [...] ng time of the network without affecting its efficiency. There were evaluated different tools that work with ANN and were selected Weka algorithm for extracting the network and the Platform for Distributed Task Tarenal to distribute the training of multilayer perceptron. Finally, it was developed a training algorithm for local and distributed the MLP with the possibility of varying transfer functions. It was shown that depending on the training sample, the change of transfer functions can yield results much more efficient than those obtained with the classic sigmoid function with increased g-media between 4.5 and 17 %. Moreover, it was found that with distributed training can be achieved eventually, better results than those achieved in the local environment.

Yuleidys, Mejías César; Ramón, Carrasco Velar; Isbel, Ochoa Izquierdo; Edel, Moreno Lemus.

86

Scientific Electronic Library Online (English)

Full Text Available SciELO Cuba | Language: Spanish Abstract in spanish El perceptrón multicapa (PMC) figura dentro de los tipos de redes neuronales artificiales (RNA) con resultados útiles en los estudios de relación estructura-actividad. Dado que los volúmenes de datos en proyectos de Bioinformática son eventualmente grandes, se propuso evaluar algoritmos para acortar [...] el tiempo de entrenamiento de la red sin afectar su eficiencia. Se desarrolló un algoritmo para el entrenamiento local y distribuido del PMC con la posibilidad de variar las funciones de transferencias para lo cual se utilizaron el Weka y la Plataforma de Tareas Distribuidas Tarenal para distribuir el entrenamiento del perceptrón multicapa. Se demostró que en dependencia de la muestra de entrenamiento, la variación de las funciones de transferencia pueden reportar resultados mucho más eficientes que los obtenidos con la clásica función Sigmoidal, con incremento de la g-media entre el 4.5 y el 17 %. Se encontró además que en los entrenamientos distribuidos es posible alcanzar eventualmente mejores resultados que los logrados en ambiente local. Abstract in english The multilayer perceptron (PMC) ranks among the types of artificial neural networks (ANN), which has provided better results in studies of structure-activity relationship. As the data volumes in Bioinformatics' projects are eventually big, it was proposed to evaluate algorithms to shorten the traini [...] ng time of the network without affecting its efficiency. There were evaluated different tools that work with ANN and were selected Weka algorithm for extracting the network and the Platform for Distributed Task Tarenal to distribute the training of multilayer perceptron. Finally, it was developed a training algorithm for local and distributed the MLP with the possibility of varying transfer functions. It was shown that depending on the training sample, the change of transfer functions can yield results much more efficient than those obtained with the classic sigmoid function with increased g-media between 4.5 and 17 %. Moreover, it was found that with distributed training can be achieved eventually, better results than those achieved in the local environment.

Yuleidys, Mejías César; Ramón, Carrasco Velar; Isbel, Ochoa Izquierdo; Edel, Moreno Lemus.

2013-12-01

87

Evolutionary Learning Algorithm for Multi-layer Morphological Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available Morphological Neural Network (MNN is a novel and important neural network and it has many applications such as image processing and pattern recognition. It makes sense to research the learning algorithm of MNN and its application. A method based on genetic algorithm is presented to train and implement multi-layer morphological neural network in this study. The algorithm calculates the weights and biases of morphological neural network and the genetic algorithm automatically acquire the learning rate. After that, the trained morphological neural network is applied to image restoration. The image restoration simulation and a comparison with the median filter are shown in the end. It shows that the morphological neural network is a quite good method applied to image restoration.

He Chunmei

2013-01-01

88

Local linear perceptrons for classification.

A structure composed of local linear perceptrons for approximating global class discriminants is investigated. Such local linear models may be combined in a cooperative or competitive way. In the cooperative model, a weighted sum of the outputs of the local perceptrons is computed where the weight is a function of the distance between the input and the position of the local perceptron. In the competitive model, the cost function dictates a mixture model where only one of the local perceptrons give output. Learning of the local models' positions and the linear mappings they implement are coupled and both supervised. We show that this is preferable to the uncoupled case where the positions are trained in an unsupervised manner before the separate, supervised training of mappings. We use goodness criteria based on the cross-entropy and give learning equations for both the cooperative and competitive cases. The coupled and uncoupled versions of cooperative and competitive approaches are compared among themselves and with multilayer perceptrons of sigmoidal hidden units and radial basis functions (RBFs) of Gaussian units on the application of recognition of handwritten digits. The criteria of comparison are the generalization accuracy, learning time, and the number of free parameters. We conclude that even on such a high-dimensional problem, such local models are promising. They generalize much better than RBF's and use much less memory. When compared with multilayer perceptrons, we note that local models learn much faster and generalize as well and sometimes better with comparable number of parameters. PMID:18263476

Alpaydin, E; Jordan, M I

1996-01-01

89

SOM-MLP Multi-Layered Neural Network with False-Alarming Nodes for Large Scale Pattern Recognition

Energy Technology Data Exchange (ETDEWEB)

In this paper, an SOM-MLP multi-layered neural network was studied for the large-scale pattern recognition problem such as the multilingual character recognition. The multi-layered neural network is made of the preclassification and the fine recognition modes. We constructed clusters for the preclassification mode using self-organizing map (SOM) learning and performed modifying steps for reducing the number of clusters. The clusters contain patterns that have the similar characteristics. We adopted the multi-layer perceptron(MLP) networks to the corresponding clusters for the fine recognition mode. And we proposed the use of false-alarming nodes in output layer of the MLP network, which could be constructed on error-prone negative examples quite similar to the patterns of the selected cluster but actually belonging to different nearby clusters through SOM`s topology-preserving mapping. The proposed system could be successfully adopted for recognizing the large number of printed Korean/Chinese characters database as well as IRIS database. (author). 14 refs., 13 figs., 2 tabs.

Kang, B.S.; Lim, K.T.; Chien, S.I. [Kyungpook National University, Taegu (Korea, Republic of)

1999-04-01

90

The synaptic morphological perceptron

In recent years, several researchers have constructed novel neural network models based on lattice algebra. Because of computational similarities to operations in the system of image morphology, these models are often called morphological neural networks. One neural model that has been successfully applied to many pattern recognition problems is the single-layer morphological perceptron with dendritic structure (SLMP). In this model, the fundamental computations are performed at dendrites connected to the body of a single neuron. Current training algorithms for the SLMP work by enclosing the target patterns in a set of hyperboxes orthogonal to the axes of the data space. This work introduces an alternate model of the SLMP, dubbed the synaptic morphological perceptron (SMP). In this model, each dendrite has one or more synapses that receive connections from inputs. The SMP can learn any region of space determined by an arbitrary configuration of hyperplanes, and is not restricted to forming hyperboxes during training. Thus, it represents a more general form of the morphological perceptron than previous architectures.

Myers, Daniel S.

2006-08-01

91

Digital Repository Infrastructure Vision for European Research (DRIVER)

This paper focuses on the segmentation of printed Bangla characters for efficient recognition of the characters. The segmentation of characters is an important step in the process of character recognitions because it allows the system to classify the characters more accurately and quickly. The system takes the scanned image file of the printed document as its input. A structural feature extraction method is used to extract the feature. In this case, each individual Bangla character is convert...

Md. Musfique Anwar; Nasrin Sultana Shume; Moniruzzaman, P. K. M.; Md. Al-Amin Bhuiyan

2010-01-01

92

Directory of Open Access Journals (Sweden)

Full Text Available Especially for the last decade, the neural network models have been applied to solve financial problems like portfolio construction and stock market forecasting. Among the alternative neural network models, the multilayer perceptron models are expected to be effective and widely applied in financial forecasting. This study examines the forecasting power multilayer perceptron models for daily and sessional returns of ISE-100 index. The findings imply that the multilayer perceptron models presented promising performance in forecasting the ISE-100 index returns. However, further emphasis should be placed on different input variables and model architectures in order to improve the forecasting performances.

Emin AVCI

2007-06-01

93

A novel single neuron perceptron with universal approximation and XOR computation properties.

We propose a biologically motivated brain-inspired single neuron perceptron (SNP) with universal approximation and XOR computation properties. This computational model extends the input pattern and is based on the excitatory and inhibitory learning rules inspired from neural connections in the human brain's nervous system. The resulting architecture of SNP can be trained by supervised excitatory and inhibitory online learning rules. The main features of proposed single layer perceptron are universal approximation property and low computational complexity. The method is tested on 6 UCI (University of California, Irvine) pattern recognition and classification datasets. Various comparisons with multilayer perceptron (MLP) with gradient decent backpropagation (GDBP) learning algorithm indicate the superiority of the approach in terms of higher accuracy, lower time, and spatial complexity, as well as faster training. Hence, we believe the proposed approach can be generally applicable to various problems such as in pattern recognition and classification. PMID:24868200

Lotfi, Ehsan; Akbarzadeh-T, M-R

2014-01-01

94

Predicting shifts in generalization gradients with perceptrons.

Perceptron models have been used extensively to model perceptual learning and the effects of discrimination training on generalization, as well as to explore natural classification mechanisms. Here, we assess the ability of existing models to account for the time course of generalization shifts that occur when individuals learn to distinguish sounds. A set of simulations demonstrates that commonly used single-layer and multilayer perceptron networks do not predict transitory shifts in generalization over the course of training but that such dynamics can be accounted for when the output functions of these networks are modified to mimic the properties of cortical tuning curves. The simulations further suggest that prudent selection of stimuli and training criteria can allow for more precise predictions of learning-related shifts in generalization gradients in behavioral experiments. In particular, the simulations predict that individuals will show maximal peak shift after different numbers of trials, that easier contrasts will lead to slower development of shifted peaks, and that whether generalization shifts persist or dissipate will depend on which stimulus dimensions individuals use to distinguish stimuli and how those dimensions are neurally encoded. PMID:21983938

Wisniewski, Matthew G; Radell, Milen L; Guillette, Lauren M; Sturdy, Christopher B; Mercado, Eduardo

2012-06-01

95

The use of artificial neural networks for residential buildings conceptual cost estimation

Accurate cost estimation in the early phase of the building's design process is of key importance for a project's success. Both underestimation and overestimation may lead to projects failure in terms of costs. The paper presents synthetically some research results on the use of neural networks for conceptual cost estimation of residential buildings. In the course of the research the author focused on regression models binding together the basic information about residential buildings available in the early stage of design and construction cost. Application of different neural networks types was analysed (multilayer perceptron, multilayer perceptron with data compression based on principal component analysis and radial basis function networks). Due to the research results, multilayer perceptron networks proved to be the best neural network type for the problem solution. The research results indicate that a neural approach may be an interesting alternative for the traditional methods of conceptual cost estimation in construction projects.

Juszczyk, Micha?

2013-10-01

96

Directory of Open Access Journals (Sweden)

Full Text Available En este trabajo se realizó un estudio estadístico de variables físico químicas asociadas al fenómeno de contaminación ambiental, en particular concentración media mensual de SO2 , medidas en la ciudad Salta Capital, Argentina, simultáneamente a concentraciones de NO2 y O3 . Las series bajo estudio presentaban comportamientos dinámicos no lineales, datos atípicos y cambios estructurales, lo que hizo imposible modelarlas con tipologías econométricas tradiciones (AR, MA, ARMA, ARIMA, entre otras. Una solución eficiente que se encontró, hace uso de la teoría de los perceptrones multicapa. Mediante el modelo estructural de series de tiempo, esta solución se presenta como un proceso matemático iterativo que permite obtener un modelado final el cual tiene una muy alta confiabilidad (95%, para realizar pronoósticos a futuro sobre el comportamiento de la variable estudiada.In this paper a statistical study of phisical-chemistry variables connected with enviroment pollution, specifically SO2 monthly average concentration, measured in Salta Capital city, Argentina, together with NO2 and O3 concentrations, was made. Time series under study shown non linear dinamic behaviour, outliers and structural changes. Due to these it was impossible to use typical econometric typologies (AR, MA, ARMA, ARIMA, among others. An effective solution which uses multistep perceptrons theory was found. By using structural time series modelling, this solution is presented by an iterative mathematical process that allows us to obtain a final model with a high confidence level (95% in order to do the forecasting step on the studied variable.

Haydeé Elena Musso

2013-01-01

97

Artificial neural networks are a computational paradigm inspired by biological neural systems. By modeling neural networks to a certain degree after their counterparts in nature, it is hoped that they can capture those aspects of biological neural systems that allow them to outperform more conventional processing systems in tasks such as motor control and pattern recognition. A brief overview of neural networks is provided in Chapter 1, concentrating on those aspects pertinent to the remainder of this thesis. The application of neural networks to control is examined in Chapter 2. A general control system can be divided into feedforward and feedback components. Specifically, the use of neural networks in learning to generate the feedforward control signal for unknown, potentially nonlinear, plants is examined. A class of learning algorithms applicable to feedforward networks is developed, and their use in learning to control a simulated two-link robotic manipulator is studied. An optoelectronic implementation of a multilayer feedforward neural network, with binary weights and connections, is described in the final part of this thesis. The neurons and connections are implemented electronically on a custom VLSI chip. The pattern and strength of the connections is controlled, through photodetectors placed in the connections, by a pattern of light illuminating the chip. This pattern is read out, in parallel, from an optical disk. Issues concerning parallel readout of information from optical disks are discussed in Chapter 3, while Chapter 4 contains a description of both the design of the Optoelectronic Neural Network Chip (ONNC) and experiments involving the optical disk and neural network chip.

Yamamura, Alan Akihiro

1992-09-01

98

Scientific Electronic Library Online (English)

Full Text Available SciELO Costa Rica | Language: Spanish Abstract in spanish En este trabajo se realizó un estudio estadístico de variables físico químicas asociadas al fenómeno de contaminación ambiental, en particular concentración media mensual de SO2 , medidas en la ciudad Salta Capital, Argentina, simultáneamente a concentraciones de NO2 y O3 . Las series bajo estudio p [...] resentaban comportamientos dinámicos no lineales, datos atípicos y cambios estructurales, lo que hizo imposible modelarlas con tipologías econométricas tradiciones (AR, MA, ARMA, ARIMA, entre otras). Una solución eficiente que se encontró, hace uso de la teoría de los perceptrones multicapa. Mediante el modelo estructural de series de tiempo, esta solución se presenta como un proceso matemático iterativo que permite obtener un modelado final el cual tiene una muy alta confiabilidad (95%), para realizar pronoósticos a futuro sobre el comportamiento de la variable estudiada. Abstract in english In this paper a statistical study of phisical-chemistry variables connected with enviroment pollution, specifically SO2 monthly average concentration, measured in Salta Capital city, Argentina, together with NO2 and O3 concentrations, was made. Time series under study shown non linear dinamic behavi [...] our, outliers and structural changes. Due to these it was impossible to use typical econometric typologies (AR, MA, ARMA, ARIMA, among others). An effective solution which uses multistep perceptrons theory was found. By using structural time series modelling, this solution is presented by an iterative mathematical process that allows us to obtain a final model with a high confidence level (95%) in order to do the forecasting step on the studied variable.

Haydeé Elena, Musso; Orlando José, Ávila Blas.

2013-01-01

99

Scientific Electronic Library Online (English)

Full Text Available SciELO Costa Rica | Language: Spanish Abstract in spanish En este trabajo se realizó un estudio estadístico de variables físico químicas asociadas al fenómeno de contaminación ambiental, en particular concentración media mensual de SO2 , medidas en la ciudad Salta Capital, Argentina, simultáneamente a concentraciones de NO2 y O3 . Las series bajo estudio p [...] resentaban comportamientos dinámicos no lineales, datos atípicos y cambios estructurales, lo que hizo imposible modelarlas con tipologías econométricas tradiciones (AR, MA, ARMA, ARIMA, entre otras). Una solución eficiente que se encontró, hace uso de la teoría de los perceptrones multicapa. Mediante el modelo estructural de series de tiempo, esta solución se presenta como un proceso matemático iterativo que permite obtener un modelado final el cual tiene una muy alta confiabilidad (95%), para realizar pronoósticos a futuro sobre el comportamiento de la variable estudiada. Abstract in english In this paper a statistical study of phisical-chemistry variables connected with enviroment pollution, specifically SO2 monthly average concentration, measured in Salta Capital city, Argentina, together with NO2 and O3 concentrations, was made. Time series under study shown non linear dinamic behavi [...] our, outliers and structural changes. Due to these it was impossible to use typical econometric typologies (AR, MA, ARMA, ARIMA, among others). An effective solution which uses multistep perceptrons theory was found. By using structural time series modelling, this solution is presented by an iterative mathematical process that allows us to obtain a final model with a high confidence level (95%) in order to do the forecasting step on the studied variable.

Haydeé Elena, Musso; Orlando José, Ávila Blas.

100

Energy Technology Data Exchange (ETDEWEB)

A modularly-structured neural network model is considered. Each module, which we call a 'cell', consists of two parts: a Hopfield neural network model and a multilayered perceptron. An array of such cells is used to simulate the Rule 110 cellular automaton with high accuracy even when all the units of neural networks are replaced by stochastic binary ones. We also find that noise not only degrades but also facilitates computation if the outputs of multilayered perceptrons are below the threshold required to update the states of the cells, which is a stochastic resonance in computation.

Oku, Makito, E-mail: oku@sat.t.u-tokyo.ac.j [Department of Mathematical Informatics, Graduate School of Information Science and Technology, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 (Japan); Aihara, Kazuyuki [Institute of Industrial Science, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505 (Japan); Department of Mathematical Informatics, Graduate School of Information Science and Technology, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 (Japan)

2010-11-01

101

International Nuclear Information System (INIS)

A modularly-structured neural network model is considered. Each module, which we call a 'cell', consists of two parts: a Hopfield neural network model and a multilayered perceptron. An array of such cells is used to simulate the Rule 110 cellular automaton with high accuracy even when all the units of neural networks are replaced by stochastic binary ones. We also find that noise not only degrades but also facilitates computation if the outputs of multilayered perceptrons are below the threshold required to update the states of the cells, which is a stochastic resonance in computation.

102

Perceptron with one layer based on optical devices

The perceptron is useful to be used in different forms and implemented into different technologies for could study of limits and development directions of neural networks in respectively technologies. In this paper the authors present, from theoretical point of view one model of perceptron with a single layer achieved with optoelectronic and optic devices. The showed perceptron has more advantage such as: its threshold can be modified, the type of inputs can be modified from excitatory to inhibitory and vice versa etc.

Degeratu, Vasile; Degeratu, Stefania; Schiopu, Paul

2005-08-01

103

Training trajectories by continuous recurrent multilayer networks.

This paper addresses the problem of training trajectories by means of continuous recurrent neural networks whose feedforward parts are multilayer perceptrons. Such networks can approximate a general nonlinear dynamic system with arbitrary accuracy. The learning process is transformed into an optimal control framework where the weights are the controls to be determined. A training algorithm based upon a variational formulation of Pontryagin's maximum principle is proposed for such networks. Computer examples demonstrating the efficiency of the given approach are also presented. PMID:18244431

Leistritz, L; Galicki, M; Witte, H; Kochs, E

2002-01-01

104

Multilayer neural-net robot controller with guaranteed tracking performance.

A multilayer neural-net (NN) controller for a general serial-link rigid robot arm is developed. The structure of the NN controller is derived using a filtered error/passivity approach. No off-line learning phase is needed for the proposed NN controller and the weights are easily initialized. The nonlinear nature of the NN, plus NN functional reconstruction inaccuracies and robot disturbances, mean that the standard delta rule using backpropagation tuning does not suffice for closed-loop dynamic control. Novel online weight tuning algorithms, including correction terms to the delta rule plus an added robust signal, guarantee bounded tracking errors as well as bounded NN weights. Specific bounds are determined, and the tracking error bound can be made arbitrarily small by increasing a certain feedback gain. The correction terms involve a second-order forward-propagated wave in the backpropagation network. New NN properties including the notions of a passive NN, a dissipative NN, and a robust NN are introduced. PMID:18255592

Lewis, F L; Yegildirek, A; Liu, K

1996-01-01

105

Learning by a population of perceptrons

Learning by examples of a population of neural networks is studied in a statistical physics framework. A population of single-layer perceptrons learns from a two-layer neural network. Each member is trained independently either from the same or from different example sets. The outputs of multiple networks are combined by majority vote. We calculate the generalization curve of the group decision of the perceptrons with both discrete and continuous weights. We find an interesting nonmonotonic learning curve for the case of discrete weights, indicating that majority vote shows optimal performance when the size of the example set is finite.

Kang, Kukjin; Oh, Jong-Hoon; Kwon, Chulan

1997-03-01

106

Discrete Orthogonal Transforms and Neural Networks for Image Interpolation

Directory of Open Access Journals (Sweden)

Full Text Available In this contribution we present transform and neural network approaches to the interpolation of images. From transform point of view, the principles from [1] are modified for 1st and 2nd order interpolation. We present several new interpolation discrete orthogonal transforms. From neural network point of view, we present interpolation possibilities of multilayer perceptrons. We use various configurations of neural networks for 1st and 2nd order interpolation. The results are compared by means of tables.

J. Polec

1999-09-01

107

In this study, a hepatitis disease diagnosis study was realized using neural network structure. For this purpose, a multilayer neural network structure was used. Levenberg-Marquardt algorithm was used as training algorithm for the weights update of the neural network. The results of the study were compared with the results of the previous studies reported focusing on hepatitis disease diagnosis and using same UCI machine learning database. We obtained a classification accuracy of 91.87% via tenfold cross validation. PMID:20703548

Bascil, M Serdar; Temurtas, Feyzullah

2011-06-01

108

Function Approximation Performance of Fuzzy Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available In this paper we propose a Multilayer Perceptron Neural Network (MLP NNconsisting of fuzzy flip-flop neurons based on various fuzzy operations applied in order toapproximate a real-life application, two input trigonometric functions, and two and sixdimensional benchmark problems. The Bacterial Memetic Algorithm with ModifiedOperator Execution Order algorithm (BMAM is proposed for Fuzzy Neural Networks(FNN training. The simulation results showed that various FNN types delivered very goodfunction approximation results.

László Gál

2010-11-01

109

A New Filter Design Method for Disturbed Multilayer Hopfield Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available This paper investigates the passivity based filtering problem for multilayer Hopfield neural networks with external disturbance. A new passivity based filter design method for multilayer Hopfield neural networks is developed to ensure that the filtering error system is exponentially stable and passive from the external disturbance vector to the output error vector. The unknown gain matrix is obtained by solving a linear matrix inequality (LMI, which can be easily facilitated by using some standard numerical packages. An illustrative example is given to demonstrate the effectiveness of the proposed filter.

AHN, C. K.

2011-05-01

110

ESTIMATION OF INPUT IMPEDANCE OF MICROSTRIP PATCH ANTENNA USING FUZZY NEURAL NETWORK

Directory of Open Access Journals (Sweden)

Full Text Available The paper presents the use of fuzzy neural network (FNN as a fast and better technique for the determination of input impedance of coaxial feed rectangular microstrip antenna. The fuzzy parameter ensures better performance as compared to three layer multilayered perceptron feed forward back propagation artificial neural network (MLPFFBP ANN and radial basis function artificial neural network (RBF ANN in the determination of input impedance of the coaxial feed microstrip antenna.

VANDANA VIKAS THAKARE

2010-10-01

111

Digital Repository Infrastructure Vision for European Research (DRIVER)

The main goal of the presented work is to analyse the performance of the Multi-Layer Perceptron (MLP) neural network for flow regime classification based on sets of simulated Electrical Capacitance Tomography (ECT) data. Normalised ECT data have been used to separately train several MLPs employing various co...

Khursiah Zainal-Mokhtar; Junita Mohamad-Saleh; Hafizah Talib; Najwan Osman-Ali

2009-01-01

112

Classification of Magneto-Optic Images using Neural Networks

A real time imaging system with a neural network classifier has been incorporated on a Macintosh computer in conjunction with an MOI system. This system images rivets on aircraft aluminium structures using eddy currents and magnetic imaging. Moment invariant functions from the image of a rivet is used to train a multilayer perceptron neural network to classify the rivets as good or bad (rivets with cracks).

Nath, Shridhar; Wincheski, Buzz; Fulton, Jim; Namkung, Min

1994-01-01

113

Bearing Fault Detection Using Artificial Neural Networks and Genetic Algorithm

Digital Repository Infrastructure Vision for European Research (DRIVER)

A study is presented to compare the performance of bearing fault detection using three types of artificial neural networks (ANNs), namely, multilayer perceptron (MLP), radial basis function (RBF) network, and probabilistic neural network (PNN). The time domain vibration signals of a rotating machine with normal and defective bearings are processed for feature extraction. The extracted features from original and preprocessed signals are used as inputs to all three ANN classifiers: MLP,...

Samanta B; Al-Balushi Khamis R; Al-Araimi Saeed A

2004-01-01

114

Neural Network Revisited: Perception on Modified Poincare Map of Financial Time Series Data

Artificial Neural Network Model for prediction of time-series data is revisited on analysis of the Indonesian stock-exchange data. We introduce the use of Multi-Layer Perceptron to percept the modified Poincare-map of the given financial time-series data. The modified Poincare-map is believed to become the pattern of the data that transforms the data in time-t versus the data in time-t+1 graphically. We built the Multi-Layer Perceptron to percept and demonstrate predicting the data on specific stock-exchange in Indonesia.

Situngkir, H; Situngkir, Hokky; Surya, Yohanes

2004-01-01

115

Long-term load forecasting via a hierarchical neural model with time integrators

Energy Technology Data Exchange (ETDEWEB)

A novel hierarchical hybrid neural model to the problem of long-term load forecasting is proposed in this paper. The neural model is made up of two self-organizing map nets - one on top of the other -, and a single-layer perceptron. It has application into domains which require time series analysis. The model is compared to a multilayer perceptron. Both the hierarchical and the multilayer perceptron models are trained and assessed on load data extracted from a North-American electric utility. They are required to predict either once every week or once every month the electric peak-load and mean-load during the next two years. The results are presented and evaluated in the paper. (author)

Carpinteiro, Otavio A.S.; Pinheiro, Carlos A.M.; Moreira, Edmilson M. [Research Group on Computer Networks and Software Engineering, Federal University of Itajuba, Av. BPS 1303, Itajuba, MG, 37500-903 (Brazil); Leme, Rafael C.; de Souza, Antonio C. Zambroni [Research Group on Electrical Systems Engineering, Federal University of Itajuba, Av. BPS 1303, Itajuba, MG, 37500-903 (Brazil)

2007-03-15

116

Distinction of The Authors of Texts Using Multilayered Feedforward Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available his paper proposes a means of using a multilayered feedforward neural network to identify the author of a text. The network has to be trained where multilayer feedforward neural network as a powerful scheme for learning complex input-output mapping have been used in learning of the average number of words and average characters of words in a paragraphs of an author. The resulting training information we get will be used to identify the texts written by authors. The computational complexity is solved by dividing it into a number of computationally simple tasks where the input space is divided into a set of subspaces and then combining the solutions to those tasks. By this, we have been able to successfully distinguish the books authored by Leo Tolstoy, from the ones authored by George Orwell and Boris Pasternak.

Suvad Selman

2012-03-01

117

Multi-layer holographic bifurcative neural network system for real-time adaptive EOS data analysis

Optical data processing techniques have the inherent advantage of high data throughout, low weight and low power requirements. These features are particularly desirable for onboard spacecraft in-situ real-time data analysis and data compression applications. the proposed multi-layer optical holographic neural net pattern recognition technique will utilize the nonlinear photorefractive devices for real-time adaptive learning to classify input data content and recognize unexpected features. Information can be stored either in analog or digital form in a nonlinear photofractive device. The recording can be accomplished in time scales ranging from milliseconds to microseconds. When a system consisting of these devices is organized in a multi-layer structure, a feedforward neural net with bifurcating data classification capability is formed. The interdisciplinary research will involve the collaboration with top digital computer architecture experts at the University of Southern California.

Liu, Hua-Kuang; Huang, K. S.; Diep, J.

1993-01-01

118

Comparision of Neural Algorithms for Funchtion Approximation

Directory of Open Access Journals (Sweden)

Full Text Available In this work, various neural network algorithms have been compared for function approximation problems. Multilayer Perceptron (MLP structure with standard back propagation, MLP with fast back propagation (adaptive learning and momentum term added, MLP with Levenberg-Marquardt learning algorithms, Radial Basis Function (RBF network structure trained by OLS algorithm and Conic Section Function Neural Network (CSFNN with adaptive learning have been investigated for various functions. Results showed that the neural algorithms can be used for functional estimation as an alternative to classical methods.

Lale Ozyilmaz

2002-01-01

119

Digital Repository Infrastructure Vision for European Research (DRIVER)

In this paper, Multi-Layer Perceptron and Radial-Basis Function Neural Networks, along with the Nearest Neighbour approach and linear regression are utilized for flash-flood forecasting in the mountainous Nysa Klodzka river catchment. It turned out that the Radial-Basis Function Neural Network is the best model for 3- and 6-h lead time prediction and the only reliable one for 9-h lead time forecasting for the largest flood used as a test case.

Piotrowski, A.; Napio?rkowski, J. J.; Ski, P. M. Rowi Amp X.

2006-01-01

120

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, Multi-Layer Perceptron and Radial-Basis Function Neural Networks, along with the Nearest Neighbour approach and linear regression are utilized for flash-flood forecasting in the mountainous Nysa Klodzka river catchment. It turned out that the Radial-Basis Function Neural Network is the best model for 3- and 6-h lead time prediction and the only reliable one for 9-h lead time forecasting for the largest flood used as a test case.

A. Piotrowski

2006-01-01

121

In this paper, Multi-Layer Perceptron and Radial-Basis Function Neural Networks, along with the Nearest Neighbour approach and linear regression are utilized for flash-flood forecasting in the mountainous Nysa Klodzka river catchment. It turned out that the Radial-Basis Function Neural Network is the best model for 3- and 6-h lead time prediction and the only reliable one for 9-h lead time forecasting for the largest flood used as a test case.

Piotrowski, A.; Napiórkowski, J. J.; Rowi?ski, P. M.

2006-08-01

122

Weight decay induced phase transitions in multilayer neural networks

We investigate layered neural networks with differentiable activation function and student vectors without normalization constraint by means of equilibrium statistical physics. We consider the learning of perfectly realizable rules and find that the length of student vectors becomes infinite, unless a proper weight decay term is added to the energy. Then, the system undergoes a first order phase transition between states with very long student vectors and states where the lengths are comparable to those of the teacher vectors. Additionally in both configurations there is a phase transition between a specialized and an unspecialized phase. An anti-specialized phase with long student vectors exists in networks with a small number of hidden units.

Ahr, M; Schlösser, E

1999-01-01

123

Scientific Electronic Library Online (English)

Full Text Available SciELO Colombia | Language: Spanish Abstract in spanish Este artículo presenta los resultados obtenidos al desarrollar una metodología para la detección de 5 tipos de latidos cardiacos (Normal (N), Bloqueo de Rama Derecha (RBBB), Bloqueo de Rama Izquierda (LBBB), Contracción Auricular Prematura (APC) y Contracción Ventricular Prematura (PVC)) utilizando [...] la transformada por paquetes Wavelet de manera no adaptativa en la extracción de características de las señales cardiacas, empleando la función Shanon para cálculo de la entropía y adicionando una fase de identificación de nodos por cada tipo de señal cardiaca en el árbol Wavelet. La utilización de la transformada por paquetes Wavelet permite acceder a información obtenida de la descomposición tanto de baja como de alta frecuencia proporcionando un análisis más integral que el logrado con la transformada Wavelet discreta. Se evaluaron Wavelets madre de las familias Daubechies, Symlet 5 y Biortogonal inversa; que fueron resultado de una investigación previa en que se identificaron las Wavelet madre que mayor entropía presentaban con las señales cardiacas. Con la modalidad no adaptativa se reduce el costo computacional al utilizar los paquetes Wavelet, coste que representa la mayor desventaja de esta transformada, dando validez a la investigación realizada. Para la clasificación de los patrones cardiacos se emplearon las máquinas de soporte vectorial y el perceptrón multicapa. Con las máquinas de soporte vectorial empleando kernel de función de base radial, se logró un error de clasificación del 2,57 %. Abstract in english This paper presents the results obtained by developing a methodology to detect 5 types of heartbeats (Normal (N), Right bundle branch block (RBBB), Left bundle branch block (LBBB), Premature atrial contraction (APC) and Premature ventricular contraction (PVC)), using Wavelet transform packets with n [...] on-adaptative mode applied on features extraction from heartbeats. It was used the Shannon function to calculate the entropy and It was added an identification nodes stage per every type of cardiac signal in the Wavelet tree. The using of Wavelet packets transform allows the access to information which results of decomposition of low and high frecuency, giving providing a more integral analysis than achieved by the discrete Wavelet transform. Three families of mother Wavelet were evaluated on transformation: Daubechies, Symlet and Reverse Biorthogonal, which were results from a previous research in that were identified the mother Wavelet that had higher entropy with the cardiac signals. With non-adaptive mode, the computational cost is reduced when Wavelet packets are used; this cost represents the most marked disadvantage from the transform. To classify the heartbeats were used Support Vector Machines and Multilayer Perceptron. The best classification error was achieved employing Support Vector Machine and a radial basis function; it was 2.57 %.

Alejandro J., Orozco-Naranjo; Pablo A., Muñoz-Gutiérrez.

2013-12-30

124

Modeling of an industrial drying process by artificial neural networks

Directory of Open Access Journals (Sweden)

Full Text Available A suitable method is needed to solve the nonquality problem in the grated coconut industry due to the poor control of product humidity during the process. In this study the possibility of using an artificial neural network (ANN, precisely a Multilayer Perceptron, for modeling the drying step of the production of grated coconut process is highlighted. Drying must confer to the product a final moisture of 3%. Unfortunately, under industrial conditions, this moisture varies from 1.9 to 4.8 %. In order to control this parameter and consequently reduce the proportion of the product that does not meet the humidity specification, a 9-4-1 neural network architecture was established using data gathered from an industrial plant. This Multilayer Perceptron can satisfactorily model the process with less bias, ranging from -0.35 to 0.34%, and can reduce the rate of rejected products from 92% to 3% during the first cycle of drying.

E. Assidjo

2008-09-01

125

Use of multilayer feedforward neural nets as a display method for multidimensional distributions.

We present a new method based on multilayer feedforward neural nets for displaying an n-dimensional distribution in a projected space of 1, 2 or 3 dimensions. A fully nonlinear net with several hidden layers is used. Efficient learning is achieved using multi-seed backpropagation. As a principal component analysis (PCA), the proposed method is useful for extracting information on the structure of the data set, but unlike the PCA, the transformation between the original distribution and the projected one is not restricted to be linear. Artificial examples and a real application are presented in order to show the reliability and potential of the method. PMID:8589864

Garrido, L; Gaitan, V; Serra-Ricart, M; Calbet, X

1995-09-01

126

Handwritten Farsi Character Recognition using Artificial Neural Network

Digital Repository Infrastructure Vision for European Research (DRIVER)

Neural Networks are being used for character recognition from last many years but most of the work was confined to English character recognition. Till date, a very little work has been reported for Handwritten Farsi Character recognition. In this paper, we have made an attempt to recognize handwritten Farsi characters by using a multilayer perceptron with one hidden layer. The error backpropagation algorithm has been used to train the MLP network. In addition, an analysis ha...

Reza Gharoie Ahangar; Mohammad Farajpoor Ahangar

2009-01-01

127

Evolutionary design of neural networks for classification and regression

Digital Repository Infrastructure Vision for European Research (DRIVER)

The Multilayer Perceptrons (MLPs) are the most popular class of Neural Networks. When applying MLPs, the search for the ideal architecture is a crucial task, since it should should be complex enough to learn the input/output mapping, without overfitting the training data. Under this context, the use of Evolutionary Computation makes a promising global search approach for model selection. On the other hand, ensembles (combinations of models) have been boosting the performance of several Machin...

Rocha, Miguel; Cortez, Paulo; Neves, Jos

2005-01-01

128

Artificial Neural Network to predict mean monthly total ozone in Arosa, Switzerland

Present study deals with the mean monthly total ozone time series over Arosa, Switzerland. The study period is 1932-1971. First of all, the total ozone time series has been identified as a complex system and then Artificial Neural Networks models in the form of Multilayer Perceptron with back propagation learning have been developed. The models are Single-hidden-layer and Two-hidden-layer Perceptrons with sigmoid activation function. After sequential learning with learning rate 0.9 the peak total ozone period (February-May) concentrations of mean monthly total ozone have been predicted by the two neural net models. After training and validation, both of the models are found skillful. But, Two-hidden-layer Perceptron is found to be more adroit in predicting the mean monthly total ozone concentrations over the aforesaid period.

Chattopadhyay, S; Chattopadhyay, Surajit; Bandyopadhyay, Goutami

2006-01-01

129

Design and FPGA-implementation of multilayer neural networks with on-chip learning

International Nuclear Information System (INIS)

Artificial Neural Networks (ANN) is used in many applications in the industry because of their parallel structure, high speed, and their ability to give easy solution to complicated problems. For example identifying the orange and apple in the sorting machine with neural network is easier than using image processing techniques to do the same thing. There are different software for designing, training, and testing the ANN, but in order to use the ANN in the industry, it should be implemented on hardware outside the computer. Neural networks are artificial systems inspired on the brain's cognitive behavior, which can learn tasks with some degree of complexity, such as signal processing, diagnosis, robotics, image processing, and pattern recognition. Many applications demand a high computing power and the traditional software implementation are not sufficient.This thesis presents design and FPGA implementation of Multilayer Neural Networks with On-chip learning in re-configurable hardware. Hardware implementation of neural network algorithm is very interesting due their high performance and they can easily be made parallel. The architecture proposed herein takes advantage of distinct data paths for the forward and backward propagation stages and a pipelined adaptation of the on- line backpropagation algorithm to significantly improve the performance of the learning phase. The architecture is easily scalable and able to cope with arbitrary network sizes with the same hardware. The implementation is targeted diagnosis of the Research Reactor accidents to avoid the risk of occurrence of a nuclear accident. The proposed designed circuits are implemented using Xilinx FPGA Chip XC40150xv and occupied 73% of Chip CLBs. It achieved 10.8 ?s to take decision in the forward propagation compared with current software implemented of RPS which take 24 ms. The results show that the proposed architecture leads to significant speed up comparing to high end software solutions. On-chip learning allows on line reconstruction of ANN. Re-configure ability and parallel structure of FPGA makes it possible to accomplish this task.

130

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: Portuguese Abstract in portuguese Estimativa do perfil de concentração de clorofila, em águas naturais, a partir da radiação emergente na superfície de um corpo d'agua, com o uso de rede neural artificial do tipo Perceptron de Múltiplas Camadas. A concentração de clorofila está relacionada com os coeficientes de absorção e espalhame [...] nto via modelos bio-ópticos. O treinamento da rede é formulado como um problema de otimização, no qual a atualização das variáveis livres da rede (pesos, viés e parâmetros de cada função de ativação) é feita através do método quasi-Newton. Abstract in english In this work the average profile of chlorophyll concentration is estimated from the emitted radiation at the surface of natural waters. This is performed through the use an Artificial Neural Network of Multilayer Perceptron type to act as the inverse operator. Bio-optical models are used to correlat [...] e the chlorophyll concentration with the absorption and scattering coefficients. The network training is formulated as an optimization problem, in which the update of the free variables of network (weights, viéses and each slope of the activation functions) is performed through the quasi-Newton method.

F., Dall Cortivo; E. S., Chalhoub; H. F., Campos Velho.

2012-12-01

131

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: Portuguese Abstract in portuguese Estimativa do perfil de concentração de clorofila, em águas naturais, a partir da radiação emergente na superfície de um corpo d'agua, com o uso de rede neural artificial do tipo Perceptron de Múltiplas Camadas. A concentração de clorofila está relacionada com os coeficientes de absorção e espalhame [...] nto via modelos bio-ópticos. O treinamento da rede é formulado como um problema de otimização, no qual a atualização das variáveis livres da rede (pesos, viés e parâmetros de cada função de ativação) é feita através do método quasi-Newton. Abstract in english In this work the average profile of chlorophyll concentration is estimated from the emitted radiation at the surface of natural waters. This is performed through the use an Artificial Neural Network of Multilayer Perceptron type to act as the inverse operator. Bio-optical models are used to correlat [...] e the chlorophyll concentration with the absorption and scattering coefficients. The network training is formulated as an optimization problem, in which the update of the free variables of network (weights, viéses and each slope of the activation functions) is performed through the quasi-Newton method.

F., Dall Cortivo; E. S., Chalhoub; H. F., Campos Velho.

132

In the paper, one multi-layer BP neural network is applied to identify metastasis malignancy tumor cells in lymph node puncture image. The topology structure of the network is as following: the node number of input-layer is 9, which involves morphologic features and chroma information; the node number of first hide-layer and second hide-layer is defined respectively as 8 and 12; the node number of output-layer is 3, which is the category number of recognized objects. Experimental results show that the learning performance of multi-layer BP neural network is good, comparing with three-layer BP neural network, the recognition rate is improved, and the method can be as an assistant means to recognize lymph node metastasis malignancy tumor cells.

Kong, Ling; Liu, Chunping; Shen, Peihua; Xia, Deshen

2001-09-01

133

Directory of Open Access Journals (Sweden)

Full Text Available The ability of Multi-Layer Perceptron (MLP and Neuro-Fuzzy neural networks to classify corn seed varieties based on mixed morphological and color Features has been evaluated that would be helpful for automation of corn handling. This research was done in Islamic Azad University, Shahr-e-Rey Branch, during 2011 on 5 main corn varieties were grown in different environments of Iran. A total of 12 color features, 11 morphological features and 4 shape factors were extracted from color images of each corn kernel. Two types of neural networks contained Multilayer Perceptron (MLP and Neuro-Fuzzy were used to classify the corn seed varieties. Average classification’s accuracy of corn seed varieties were obtained 94% and 96% by MLP and Neuro-Fuzzy classifiers respectively. After feature selection by UTA algorithm, more effective features were selected to decrease the classification processing time, without any meaningful decreasing of accuracies.

Alireza Pazoki

2013-10-01

134

Neural network approximation of nonlinearity in laser nano-metrology system based on TLMI

International Nuclear Information System (INIS)

In this paper, an approach based on neural network (NN) for nonlinearity modeling in a nano-metrology system using three-longitudinal-mode laser heterodyne interferometer (TLMI) for length and displacement measurements is presented. We model nonlinearity errors that arise from elliptically and non-orthogonally polarized laser beams, rotational error in the alignment of laser head with respect to the polarizing beam splitter, rotational error in the alignment of the mixing polarizer, and unequal transmission coefficients in the polarizing beam splitter. Here we use a neural network algorithm based on the multi-layer perceptron (MLP) network. The simulation results show that multi-layer feed forward perceptron network is successfully applicable to real noisy interferometer signals.

135

Directory of Open Access Journals (Sweden)

Full Text Available Neural Networks are a set of mathematical methods and computer programs designed to simulate the information process and the knowledge acquisition of the human brain. In last years its application in chemistry is increasing significantly, due the special characteristics for model complex systems. The basic principles of two types of neural networks, the multi-layer perceptrons and radial basis functions, are introduced, as well as, a pruning approach to architecture optimization. Two analytical applications based on near infrared spectroscopy are presented, the first one for determination of nitrogen content in wheat leaves using multi-layer perceptrons networks and second one for determination of BRIX in sugar cane juices using radial basis functions networks.

Eduardo O. de Cerqueira

2001-12-01

136

The Perceptron with Dynamic Margin

The classical perceptron rule provides a varying upper bound on the maximum margin, namely the length of the current weight vector divided by the total number of updates up to that time. Requiring that the perceptron updates its internal state whenever the normalized margin of a pattern is found not to exceed a certain fraction of this dynamic upper bound we construct a new approximate maximum margin classifier called the perceptron with dynamic margin (PDM). We demonstrate that PDM converges in a finite number of steps and derive an upper bound on them. We also compare experimentally PDM with other perceptron-like algorithms and support vector machines on hard margin tasks involving linear kernels which are equivalent to 2-norm soft margin.

Panagiotakopoulos, Constantinos

2011-01-01

137

Video Traffic Prediction Using Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, we consider video stream prediction for application in services likevideo-on-demand, videoconferencing, video broadcasting, etc. The aim is to predict thevideo stream for an efficient bandwidth allocation of the video signal. Efficient predictionof traffic generated by multimedia sources is an important part of traffic and congestioncontrol procedures at the network edges. As a tool for the prediction, we use neuralnetworks – multilayer perceptron (MLP, radial basis function networks (RBF networksand backpropagation through time (BPTT neural networks. At first, we briefly introducetheoretical background of neural networks, the prediction methods and the differencebetween them. We propose also video time-series processing using moving averages.Simulation results for each type of neural network together with final comparisons arepresented. For comparison purposes, also conventional (non-neural prediction isincluded. The purpose of our work is to construct suitable neural networks for variable bitrate video prediction and evaluate them. We use video traces from [1].

Miloš Oravec

2008-10-01

138

Neural network modeling and correcting for delay-line data sets

International Nuclear Information System (INIS)

Because of the effects of the capacitance and inductance parasitized on the readout PCB in GEM detector, the output time of the delay-line PCB puts up a non-linear relationship with the position of its input signal. Based on Back Propagation algorithm, the multi-layer perceptrons neural network approximated the non-linear function and gave out accurate analyses, which is a better method for data correcting in Delay-Line readout. (authors)

139

Viscosity Calculation at Moderate Pressure for Nonpolar Gases via Neural Network

Digital Repository Infrastructure Vision for European Research (DRIVER)

A new method, based on Artificial Neural Networks (ANN) of Multi-Layer Perceptron (MLP) type, has been developed to estimate the viscosity at moderate pressure for pure nonpolar gases over a wide range of temperatures. An ANN was trained, using four physicochemical properties: Molecular weight (M), boiling point (Tb), critical Temperature (Tc) and critical Pressure (Pc) combined with absolute Temperature (T) as its inputs, to correlate and predict viscosity. A...

Bouzidi, A.; Hanini, S.; Souahi, F.; Mohammedi, B.; Touiza, M.

2007-01-01

140

A design philosophy for multi-layer neural networks with applications to robot control

A system is proposed which receives input information from many sensors that may have diverse scaling, dimension, and data representations. The proposed system tolerates sensory information with faults. The proposed self-adaptive processing technique has great promise in integrating the techniques of artificial intelligence and neural networks in an attempt to build a more intelligent computing environment. The proposed architecture can provide a detailed decision tree based on the input information, information stored in a long-term memory, and the adapted rule-based knowledge. A mathematical model for analysis will be obtained to validate the cited hypotheses. An extensive software program will be developed to simulate a typical example of pattern recognition problem. It is shown that the proposed model displays attention, expectation, spatio-temporal, and predictory behavior which are specific to the human brain. The anticipated results of this research project are: (1) creation of a new dynamic neural network structure, and (2) applications to and comparison with conventional multi-layer neural network structures. The anticipated benefits from this research are vast. The model can be used in a neuro-computer architecture as a building block which can perform complicated, nonlinear, time-varying mapping from a multitude of input excitory classes to an output or decision environment. It can be used for coordinating different sensory inputs and past experience of a dynamic system and actuating signals. The commercial applications of this project can be the creation of a special-purpose neuro-computer hardware which can be used in spatio-temporal pattern recognitions in such areas as air defense systems, e.g., target tracking, and recognition. Potential robotics-related applications are trajectory planning, inverse dynamics computations, hierarchical control, task-oriented control, and collision avoidance.

Vadiee, Nader; Jamshidi, MO

1989-01-01

141

Cosmic-ray discrimination capabilities of ?E-E silicon nuclear telescopes using neural networks

International Nuclear Information System (INIS)

An isotope classifier of cosmic-ray events collected by space detectors has been implemented using a multi-layer perceptron neural architecture. In order to handle a great number of different isotopes a modular architecture of the 'mixture of experts' type is proposed. The performance of this classifier has been tested on simulated data and has been compared with a 'classical' classifying procedure. The quantitative comparison with traditional techniques shows that the neural approach has classification performances comparable - within 1% - with that of the classical one, with efficiency of the order of 98%. A possible hardware implementation of such a kind of neural architecture in future space missions is considered

142

Complex Chebyshev-polynomial-based unified model (CCPBUM) neural networks

In this paper, we propose complex Chebyshev Polynomial Based unified model neural network for the approximation of complex- valued function. Based on this approximate transformable technique, we have derived the relationship between the single-layered neural network and multi-layered perceptron neural network. It is shown that the complex Chebyshev Polynomial Based unified model neural network can be represented as a functional link network that are based on Chebyshev polynomial. We also derived a new learning algorithm for the proposed network. It turns out that the complex Chebyshev Polynomial Based unified model neural network not only has the same capability of universal approximator, but also has faster learning speed than conventional complex feedforward/recurrent neural network.

Jeng, Jin-Tsong; Lee, Tsu-Tian

1998-03-01

143

Beam forming of ultra-wideband pulses by a complex-valued spatio-temporal multilayer neural network.

We present a neuro-beam former of ultra-wideband (UWB) pulses employing complex-valued spatio-temporal multilayer neural network, where complex-valued back propagation through time (CV-BPTT) is used as a learning algorithm. The system performance is evaluated with a UWB mono-cycle pulse. Simulation results in suppressing multiple UWB interferes and in steering to multiple desired UWB pulses, demonstrates the applicability of the proposed system. PMID:15912585

Suksmono, Andriyan Bayu; Hirose, Akira

2005-01-01

144

Prior estimation of motion using recursive perceptron with sEMG: a case of wrist angle.

Muscle activity is followed by myoelectric potentials. Prior estimation of motion by surface electromyography can be utilized to assist the physically impaired people as well as surgeon. In this paper, we proposed a real-time method for the prior estimation of motion from surface electromyography, especially in the case of wrist angle. The method was based on the recursive processing of multi-layer perceptron, which is trained quickly. A single layer perceptron calculates quasi tensional force of muscles from surface electromyography. A three-layer perceptron calculates the wrist's change in angle. In order to estimate a variety of motions properly, the perceptron was designed to estimate motion in a short time period, e.g. 1ms. Recursive processing enables the method to estimate motion in the target time period, e.g. 50ms. The results of the experiments showed statistical significance for the precedence of estimated angle to the measured one. PMID:23367118

Kuroda, Yoshihiro; Tanaka, Takeshi; Imura, Masataka; Oshiro, Osamu

2012-01-01

145

Perceptron-like computation based on biologically-inspired neurons with heterosynaptic mechanisms

Perceptrons are one of the fundamental paradigms in artificial neural networks and a key processing scheme in supervised classification tasks. However, the algorithm they provide is given in terms of unrealistically simple processing units and connections and therefore, its implementation in real neural networks is hard to be fulfilled. In this work, we present a neural circuit able to perform perceptron's computation based on realistic models of neurons and synapses. The model uses Wang-Buzsáki neurons with coupling provided by axodendritic and axoaxonic synapses (heterosynapsis). The main characteristics of the feedforward perceptron operation are conserved, which allows to combine both approaches: whereas the classical artificial system can be used to learn a particular problem, its solution can be directly implemented in this neural circuit. As a result, we propose a biologically-inspired system able to work appropriately in a wide range of frequencies and system parameters, while keeping robust to noise and error.

Kaluza, Pablo; Urdapilleta, Eugenio

2014-10-01

146

A method for optical pattern recognition which is based on the human visual system and is suitable for hardware implementation is presented. The system is composed of two stages. The first stage detects local features such as line orientation, linestops, corners, and intersections to create a feature map, which represents the number of these features and hence is invariant to position, size, and slight deformation of an input pattern. The next stage is a multilayered neural network that classifies an input pattern to one of predefined categories using the feature map. We have found a method of detecting these features in analog hardware which would considerably speed up the process of pattern recognition. The decomposition of an input pattern into lines with different orientations is done by an array of two-dimensional orientation sensors. We have built an orientation sensor which is invariant to the position, size, and contrast of an input pattern. The generation of the feature map is currently being done in software which receives its inputs from the line orientation sensor. Linestops, corners and intersections are detected after a series of convolution and thresholding operations for each orientation. The convolution operation can be mapped into hardware using a resistive grid technique. The simulation with an example of character recognition showed that the proper selection of convolution kernels and thresholds can detect local features described above and demonstrated the feasibility of a full hardware implementation of a feature detector.

Nishimura, Masatoshi; Van der Spiegel, Jan

1995-03-01

147

Breast Fine Needle Tumor Classification using Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available The purpose of this study is to develop an intelligent diagnosis system for breast cancer classification. Artificial Neural Networks and Support Vector Machines were being developed to classify the benign and malignant of breast tumor in fine needle aspiration cytology. First the features were extracted from 92 FNAC image. Then these features were presented to several neural network architectures to investigate the most suitable network model for classifying the tumor effectively. Four classification models were used namely multilayer perceptron (MLP using back-propagation algorithm, probabilistic neural networks (PNN, learning vector quantization (LVQ and support vector machine (SVM. The classification results were obtained using 10-fold cross validation. The performance of the networks was compared based on resulted error rate, correct rate, sensitivity and specificity. The method was evaluated using six different datasets including four datasets related to our work and two other benchmark datasets for comparison. The optimum network for classification of breast cancer cells was found using probabilistic neural networks. This is followed in order by support vector machine, learning vector quantization and multilayer perceptron. The results showed that the predictive ability of probabilistic neural networks and support vector machine are stronger than the others in all evaluated datasets.

Yasmeen M. George

2012-09-01

148

Chebyshev polynomials-based (CPB) unified model neural networks for function approximation

In this paper, we propose the approximate transformable technique, which includes the direct transformation and indirect transformation, to obtain a CPB unified model neural networks for feedforward/recurrent neural networks via Chebyshev polynomials approximation. Based on this approximate transformable technique, we have derived the relationship between the single-layer neural networks and multilayer perceptron neural networks. It is shown that the CPB unified model neural networks can be represented as a functional link networks that are based on Chebyshev polynomials, and these networks use the recursive least squares method with forgetting factor as learning algorithm. It turns out that the CPB unified model neural networks not only has the same capability of universal approximator, but also has faster learning speed than conventional feedforward/recurrent neural networks. Computer simulations show that the proposed method does have the capability of universal approximator in some functional approximation with considerable reduction in learning time.

Lee, Tsu-Tian; Jeng, Jin-Tsong

1997-04-01

149

Using Artificial Neural Networks for ECG Signals Denoising

Directory of Open Access Journals (Sweden)

Full Text Available The authors have investigated some potential applications of artificial neural networks in electrocardiografic (ECG signal prediction. For this, the authors used an adaptive multilayer perceptron structure to predict the signal. The proposed procedure uses an artificial neural network based learning structure to estimate the (n+1th sample from n previous samples To train and adjust the network weights, the backpropagation (BP algorithm was used. In this paper, prediction of ECG signals (as time series using multi-layer feedforward neural networks will be described. The results are evaluated through approximation error which is defined as the difference between the predicted and the original signal.The prediction procedure is carried out (simulated in MATLAB environment, using signals from MIT-BIH arrhythmia database. Preliminary results are encouraging enough to extend the proposed method for other types of data signals.

Zoltán Germán-Salló

2010-12-01

150

A Novel Technique to Image Annotation using Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available : Automatic annotation of digital pictures is a key technology for managing and retrieving images from large image collection. Traditional image semantics extraction and representation schemes were commonly divided into two categories, namely visual features and text annotations. However, visual feature scheme are difficult to extract and are often semantically inconsistent. On the other hand, the image semantics can be well represented by text annotations. It is also easier to retrieve images according to their annotations. Traditional image annotation techniques are time-consuming and requiring lots of human effort. In this paper we propose Neural Network based a novel approach to the problem of image annotation. These approaches are applied to the Image data set. Our main work is focused on the image annotation by using multilayer perceptron, which exhibits a clear-cut idea on application of multilayer perceptron with special features. MLP Algorithm helps us to discover the concealed relations between image data and annotation data, and annotate image according to such relations. By using this algorithm we can save more memory space, and in case of web applications, transferring of images and download should be fast. This paper reviews 50 image annotation systems using supervised machine learning Techniques to annotate images for image retrieval. Results obtained show that the multi layer perceptron Neural Network classifier outperforms conventional DST Technique.

Pankaj Savita

2013-03-01

151

The project aimed to produce an identification model that allows for automatic recognition of malting barley varieties. The project used computer image analysis and artificial neural networks. The authors based on the analysis of biological material selected set of features describing the physical parameters allowing the identification of varieties. Image analysis of samples of barley digital photographs allowed the extraction of the characteristics of varieties. Obtained characteristics from the images were used as learning data for artificial neural network. Trained a multilayer perceptron network is characterized by the identification abilities at the level of human abilities.

Nowakowski, K.; Boniecki, P.; Tomczak, R. J.; Kujawa, S.; Raba, B.

2012-04-01

152

Using neural networks for prediction of nuclear parameters

International Nuclear Information System (INIS)

Dating from 1943, the earliest work on artificial neural networks (ANN), when Warren Mc Cullock and Walter Pitts developed a study on the behavior of the biological neuron, with the goal of creating a mathematical model. Some other work was done until after the 80 witnessed an explosion of interest in ANNs, mainly due to advances in technology, especially microelectronics. Because ANNs are able to solve many problems such as approximation, classification, categorization, prediction and others, they have numerous applications in various areas, including nuclear. Nodal method is adopted as a tool for analyzing core parameters such as boron concentration and pin power peaks for pressurized water reactors. However, this method is extremely slow when it is necessary to perform various core evaluations, for example core reloading optimization. To overcome this difficulty, in this paper a model of Multi-layer Perceptron (MLP) artificial neural network type backpropagation will be trained to predict these values. The main objective of this work is the development of Multi-layer Perceptron (MLP) artificial neural network capable to predict, in very short time, with good accuracy, two important parameters used in the core reloading problem - Boron Concentration and Power Peaking Factor. For the training of the neural networks are provided loading patterns and nuclear data used in cycle 19 of Angra 1 nuclear power plant. Three models of networks are constructed using the same input data and providing the following outputs: 1- Boron Concentration and Power Peaking Factor, 2 - Boron Concentration and 3 - Power Peaking Factor. (author)

153

THE USE OF NEURAL NETWORK TECHNOLOGY TO MODEL SWIMMING PERFORMANCE

Directory of Open Access Journals (Sweden)

Full Text Available The aims of the present study were: to identify the factors which are able to explain the performance in the 200 meters individual medley and 400 meters front crawl events in young swimmers, to model the performance in those events using non-linear mathematic methods through artificial neural networks (multi-layer perceptrons and to assess the neural network models precision to predict the performance. A sample of 138 young swimmers (65 males and 73 females of national level was submitted to a test battery comprising four different domains: kinanthropometric evaluation, dry land functional evaluation (strength and flexibility, swimming functional evaluation (hydrodynamics, hydrostatic and bioenergetics characteristics and swimming technique evaluation. To establish a profile of the young swimmer non-linear combinations between preponderant variables for each gender and swim performance in the 200 meters medley and 400 meters font crawl events were developed. For this purpose a feed forward neural network was used (Multilayer Perceptron with three neurons in a single hidden layer. The prognosis precision of the model (error lower than 0.8% between true and estimated performances is supported by recent evidence. Therefore, we consider that the neural network tool can be a good approach in the resolution of complex problems such as performance modeling and the talent identification in swimming and, possibly, in a wide variety of sports

António José Silva

2007-03-01

154

Empirical modeling of nuclear power plants using neural networks

International Nuclear Information System (INIS)

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

155

Using neural networks for prediction of nuclear parameters

Energy Technology Data Exchange (ETDEWEB)

Dating from 1943, the earliest work on artificial neural networks (ANN), when Warren Mc Cullock and Walter Pitts developed a study on the behavior of the biological neuron, with the goal of creating a mathematical model. Some other work was done until after the 80 witnessed an explosion of interest in ANNs, mainly due to advances in technology, especially microelectronics. Because ANNs are able to solve many problems such as approximation, classification, categorization, prediction and others, they have numerous applications in various areas, including nuclear. Nodal method is adopted as a tool for analyzing core parameters such as boron concentration and pin power peaks for pressurized water reactors. However, this method is extremely slow when it is necessary to perform various core evaluations, for example core reloading optimization. To overcome this difficulty, in this paper a model of Multi-layer Perceptron (MLP) artificial neural network type backpropagation will be trained to predict these values. The main objective of this work is the development of Multi-layer Perceptron (MLP) artificial neural network capable to predict, in very short time, with good accuracy, two important parameters used in the core reloading problem - Boron Concentration and Power Peaking Factor. For the training of the neural networks are provided loading patterns and nuclear data used in cycle 19 of Angra 1 nuclear power plant. Three models of networks are constructed using the same input data and providing the following outputs: 1- Boron Concentration and Power Peaking Factor, 2 - Boron Concentration and 3 - Power Peaking Factor. (author)

Pereira Filho, Leonidas; Souto, Kelling Cabral, E-mail: leonidasmilenium@hotmail.com, E-mail: kcsouto@bol.com.br [Instituto Federal de Educacao, Ciencia e Tecnologia do Rio de Janeiro (IFRJ), Rio de Janeiro, RJ (Brazil); Machado, Marcelo Dornellas, E-mail: dornemd@eletronuclear.gov.br [Eletrobras Termonuclear S.A. (GCN.T/ELETRONUCLEAR), Rio de Janeiro, RJ (Brazil). Gerencia de Combustivel Nuclear

2013-07-01

156

Generalization in the programed teaching of a perceptron

According to a widely used model of learning and generalization in neural networks, a single neuron (perceptron) can learn from examples to imitate another neuron, called the teacher perceptron. We introduce a variant of this model in which examples within a layer of thickness 2Y around the decision surface are excluded from teaching. That restriction transmits global information about the teacher's rule. Therefore for a given number p=?N of presented examples (i.e., those outside of the layer) the generalization performance obtained by Boltzmannian learning is improved by setting Y to an optimum value Y0(?), which diverges for ?-->0 and remains nonzero while ?

Derényi, Imre; Geszti, Tamás; Györgyi, Géza

1994-10-01

157

Digital Repository Infrastructure Vision for European Research (DRIVER)

The Multi-Layer Feed-Forward Neural Network (MLFFNN) is applied in the context of river flow forecast combination, where a number of rainfall-runoff models are used simultaneously to produce an overall combined river flow forecast. The operation of the MLFFNN depends not only on its neuron configuration but also on the choice of neuron transfer function adopted, which is non-linear for the hidden and output layers. These models, each having a different structure to simulate the perceived mech...

Shamseldin, A. Y.; Nasr, A. E.; O’connor, K. M.

2002-01-01

158

In this paper, we propose a neural network model with a faster learning speed and a good approximate capability in the function approximation for solving worst-case identification of nonlinear systems H(infinity ) problems. Specifically, via the approximate transformable technique, we develop a Chebyshev Polynomials Based unified model neural network for solving the worst-case identification of nonlinear systems H(infinity ) problems. Based on this approximate transformable technique, the relationship between the single-layered neural network and multi-layered perceptron neural network is derived. It is shown that the Chebyshev Polynomials Based unified model neural network can be represented as a functional link network that is based on Chebyshev polynomials. We also derive a new learning algorithm such that the infinity norm of the transfer function from the input to the output is under a prescribed level. It turns out that the Chebyshev Polynomials Based unified model neural network not only has the same capability of universal approximator, but also has a faster learning speed than multi-layered perceptron or the recurrent neural network in the deterministic worst-case identification of nonlinear systems H(infinity ) problems.

Jeng, Jin-Tsong; Lee, Tsu-Tian

1998-03-01

159

Advances in Artificial Neural Networks – Methodological Development and Application

Directory of Open Access Journals (Sweden)

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 engineering.

Yanbo Huang

2009-08-01

160

APPLICATION OF NEURAL NETWORK ALGORITHMS FOR BPM LINEARIZATION

Energy Technology Data Exchange (ETDEWEB)

Stripline BPM sensors contain inherent non-linearities, as a result of field distortions from the pickup elements. Many methods have been devised to facilitate corrections, often employing polynomial fitting. The cost of computation makes real-time correction difficult, particulalry when integer math is utilized. The application of neural-network technology, particularly the multi-layer perceptron algorithm, is proposed as an efficient alternative for electrode linearization. A process of supervised learning is initially used to determine the weighting coefficients, which are subsequently applied to the incoming electrode data. A non-linear layer, known as an ?activation layer,? is responsible for the removal of saturation effects. Implementation of a perceptron in an FPGA-based software-defined radio (SDR) is presented, along with performance comparisons. In addition, efficient calculation of the sigmoidal activation function via the CORDIC algorithm is presented.

Musson, John C. [JLAB; Seaton, Chad [JLAB; Spata, Mike F. [JLAB; Yan, Jianxun [JLAB

2012-11-01

161

Three Methods to Speed up the Training of Feedforward and Feedback Perceptrons.

Training of artificial neural networks is normally a time consuming task due to iterative search imposed by the implicit nonlinearity of the network behaviour. In this work, three improvements to "batch-mode" offline training methods, gradient-based or gradient-free, are proposed. For nonlinear multilayer perceptrons (NMLP) with linear output layers, a method based on linear regression in the output layer is presented. For arbitrary NMLPs, an algorithm is developed for detecting "saturated" hidden nodes and re-activating them while transferring their contribution onto the bias node in the same layer. For state-feedback NMLPs with incomplete learning data in the state variables, a method is shown that interpolates the unknown state values to form an intermediate training set used for finding good initial weights for the final training with only the original training set. In addition, three conventional gradient-based training methods-steepest-descent gradient search, conjugate gradient, and Gauss-Newton-are compared mutually and with the above improvements on the same two example problems. Where conventional methods get stuck in bad local minima, saturation avoidance leads to satisfactory results, and the speed-up achieved by the two other improvements is about two orders of magnitude. PMID:12662484

Agarwal, Mukul; Stäger, Fritz

1997-11-01

162

Analysis and Prediction of Temperature using Statistical Artificial Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available This paper is about producing a prediction system by usingartificial neural methods that will forecast temperature. Thispaper is based on three objectives. First, study of temperature andgathers all knowledge regarding the weather this is particularlystudied in analysis part of the paper. Second, gather allknowledge about artificial neural network methods. Implementmultilayer perceptron neural network with gradient descent(backpropagation, BFGS, conjugate gradient training algorithm andwill analyze the performance of all. Lastly, achieve an objectiveof developing a temperature prediction system. The generalfinding is that with BFGS algorithm, with multilayer perceptronmodel perform well with less prediction error and more accuracythan gradient descent and conjugate gradient, thus used fortemperature prediction. To implement this project we make useof statistica software which provides the functionality calledstatistica artificial neural network(SANN which is used here fortemperature prediction and heavy weather software is used fordata gathering.

Parag Kadu

2012-04-01

163

Directory of Open Access Journals (Sweden)

Full Text Available The Multi-Layer Feed-Forward Neural Network (MLFFNN is applied in the context of river flow forecast combination, where a number of rainfall-runoff models are used simultaneously to produce an overall combined river flow forecast. The operation of the MLFFNN depends not only on its neuron configuration but also on the choice of neuron transfer function adopted, which is non-linear for the hidden and output layers. These models, each having a different structure to simulate the perceived mechanisms of the runoff process, utilise the information carrying capacity of the model calibration data in different ways. Hence, in a discharge forecast combination procedure, the discharge forecasts of each model provide a source of information different from that of the other models used in the combination. In the present work, the significance of the choice of the transfer function type in the overall performance of the MLFFNN, when used in the river flow forecast combination context, is investigated critically. Five neuron transfer functions are used in this investigation, namely, the logistic function, the bipolar function, the hyperbolic tangent function, the arctan function and the scaled arctan function. The results indicate that the logistic function yields the best model forecast combination performance. Keywords: River flow forecast combination, multi-layer feed-forward neural network, neuron transfer functions, rainfall-runoff models

A. Y. Shamseldin

2002-01-01

164

Neural network diagnosis of avascular necrosis from magnetic resonance images

We have explored the use of artificial neural networks to diagnose avascular necrosis (AVN) of the femoral head from magnetic resonance images. We have developed multi-layer perceptron networks, trained with conjugate gradient optimization, which diagnose AVN from single sagittal images of the femoral head with 100% accuracy on the training data and 97% accuracy on test data. These networks use only the raw image as input (with minimal preprocessing to average the images down to 32 X 32 size and to scale the input data values) and learn to extract their own features for the diagnosis decision. Various experiments with these networks are described.

Manduca, Armando; Christy, Paul S.; Ehman, Richard L.

1993-09-01

165

Prediction of total resistance coefficients using neural networks

Digital Repository Infrastructure Vision for European Research (DRIVER)

The Holtrop & Mennen method is widely used at the initial design stage of ships for estimating the resistance of the ship (Holtrop and Mennen, 1982). The Holtrop & Mennen method provide a prediction of the total resistance’s components. In this work we present a neural network model which performs the same task as the Holtrop & Mennem’s method, for two of the total resistance’s components. A multilayer perceptron has been therefore trained to learn the relationship between the input (le...

Ortigosa Barraga?n, Inma; Revilla Lo?pez, Guillermo; Garci?a Espinosa, Julio

2009-01-01

166

Energy Technology Data Exchange (ETDEWEB)

This paper aims at analyzing the main operation and design of operationally robust controllers in order to damp the electromechanics oscillations type inter area. For this we used an intelligent control technique based on artificial neural networks, where a multilayer perceptron it was implemented. We used a symmetrical test system of four generators, ten bars and nine transmission lines to verify the performance of the power system stabilizers and power oscillation damping (POD) for the FACTS devices, unified power flow controller (UPFC), designed for neural networks. The results show the superiority in the operation and control of oscillations in power systems using UPFC equipped with the POD.

Furini, M.A.; Araujo, P.B. de; Pereira, A.L.S. [Universidade Estadual Paulista (FEIS/UNESP), Ilha Solteira, SP (Brazil). Fac. de Engenharia. Dept. Engenharia Eletrica], Emails: mafurini@aluno.feis.unesp.br, percival@dee.feis.unesp.br, andspa@gmail.com

2009-07-01

167

Feature competition and domain of attraction in artificial-perceptron pattern recognizer

As we reported previously, learning of a multi-layered hard-limited perceptron can be formulated into a set of simultaneous linear inequalities. Solving these inequalities under a given training set would then allow us to achieve the goal of learning in this system. If the dimension N of the input vector is much larger than the number M of different patterns to be learned, then there is considerable freedom for the system to select a proper solution of the connection matrix. In most cases, even a single layer perceptron will do the learning satisfactorily. This paper reports the results of some theoretical and experimental studies of this one-layered, hard-limited perceptron trained under the novel, one-step, noniterative learning scheme. Particularly, the analysis of some important properties of this novel learning system, such as automatic feature competition, domain of convergence, and robustness of recognition, are discussed in detail.

Hu, Chia-Lun J.

1993-10-01

168

How to guess the inter magnetic bubble potential by using a simple perceptron ?

It is shown that magnetic bubble films behaviour can be described by using a 2D super-Ising hamiltonian. Calculated hysteresis curves and magnetic domain patterns are successfully compared with experimental results taken in literature. The reciprocal problem of finding paramaters of the super-Ising model to reproduce computed or experimental magnetic domain pictures is solved by using a perceptron neural network.

Padovani, S

2004-01-01

169

Artificial neural network application for predicting soil distribution coefficient of nickel

International Nuclear Information System (INIS)

The distribution (or partition) coefficient (Kd) is an applicable parameter for modeling contaminant and radionuclide transport as well as risk analysis. Selection of this parameter may cause significant error in predicting the impacts of contaminant migration or site-remediation options. In this regards, various models were presented to predict Kd values for different contaminants specially heavy metals and radionuclides. In this study, artificial neural network (ANN) is used to present simplified model for predicting Kd of nickel. The main objective is to develop a more accurate model with a minimal number of parameters, which can be determined experimentally or select by review of different studies. In addition, the effects of training as well as the type of the network are considered. The Kd values of Ni is strongly dependent on pH of the soil and mathematical relationships were presented between pH and Kd of nickel recently. In this study, the same database of these presented models was used to verify that neural network may be more useful tools for predicting of Kd. Two different types of ANN, multilayer perceptron and redial basis function, were used to investigate the effect of the network geometry on the results. In addition, each network was trained by 80 and 90% of the data and tested for 20 and 10% of the rest data. Then the results of the networks compared with the results of the mathematical models. Although the networks trained by 80 and 90% of the data the results show that all the networks predict with higher accuracy relative to mathematical models which were derived by 100% of data. More training of a network increases the accuracy of the network. Multilayer perceptron network used in this study predicts better than redial basis function network. - Highlights: ? Simplified models for predicting Kd of nickel presented using artificial neural networks. ? Multilayer perceptron and redial basis function used to predict Kd of nickel in soil. ? The neural networks predict with higher accuracy relative to mathematical models.

170

The Direct Kernel Perceptron (DKP) (Fernández-Delgado et al., 2010) is a very simple and fast kernel-based classifier, related to the Support Vector Machine (SVM) and to the Extreme Learning Machine (ELM) (Huang, Wang, & Lan, 2011), whose ?-coefficients are calculated directly, without any iterative training, using an analytical closed-form expression which involves only the training patterns. The DKP, which is inspired by the Direct Parallel Perceptron, (Auer et al., 2008), uses a Gaussian kernel and a linear classifier (perceptron). The weight vector of this classifier in the feature space minimizes an error measure which combines the training error and the hyperplane margin, without any tunable regularization parameter. This weight vector can be translated, using a variable change, to the ?-coefficients, and both are determined without iterative calculations. We calculate solutions using several error functions, achieving the best trade-off between accuracy and efficiency with the linear function. These solutions for the ? coefficients can be considered alternatives to the ELM with a new physical meaning in terms of error and margin: in fact, the linear and quadratic DKP are special cases of the two-class ELM when the regularization parameter C takes the values C=0 and C=?. The linear DKP is extremely efficient and much faster (over a vast collection of 42 benchmark and real-life data sets) than 12 very popular and accurate classifiers including SVM, Multi-Layer Perceptron, Adaboost, Random Forest and Bagging of RPART decision trees, Linear Discriminant Analysis, K-Nearest Neighbors, ELM, Probabilistic Neural Networks, Radial Basis Function neural networks and Generalized ART. Besides, despite its simplicity and extreme efficiency, DKP achieves higher accuracies than 7 out of 12 classifiers, exhibiting small differences with respect to the best ones (SVM, ELM, Adaboost and Random Forest), which are much slower. Thus, the DKP provides an easy and fast way to achieve classification accuracies which are not too far from the best one for a given problem. The C and Matlab code of DKP are freely available. PMID:24287336

Fernández-Delgado, Manuel; Cernadas, Eva; Barro, Senén; Ribeiro, Jorge; Neves, José

2014-02-01

171

Online learning in a chemical perceptron.

Autonomous learning implemented purely by means of a synthetic chemical system has not been previously realized. Learning promotes reusability and minimizes the system design to simple input-output specification. In this article we introduce a chemical perceptron, the first full-featured implementation of a perceptron in an artificial (simulated) chemistry. A perceptron is the simplest system capable of learning, inspired by the functioning of a biological neuron. Our artificial chemistry is deterministic and discrete-time, and follows Michaelis-Menten kinetics. We present two models, the weight-loop perceptron and the weight-race perceptron, which represent two possible strategies for a chemical implementation of linear integration and threshold. Both chemical perceptrons can successfully identify all 14 linearly separable two-input logic functions and maintain high robustness against rate-constant perturbations. We suggest that DNA strand displacement could, in principle, provide an implementation substrate for our model, allowing the chemical perceptron to perform reusable, programmable, and adaptable wet biochemical computing. PMID:23514238

Banda, Peter; Teuscher, Christof; Lakin, Matthew R

2013-01-01

172

In this paper, an optimized design of anode shape in order to achieve highest X-ray yield in a plasma focus device filled with nitrogen gas based on artificial neural networks (ANNs) is presented. Multi-layer perceptron neural network structure with the back-propagation algorithm is used for the training of the proposed model. The model has achieved good agreement with the training data and has yielded satisfactory generalization. This shows that the ANN model is an accurate and reliable approach to predict the highest X-ray yield in plasma focus devices.

Hayati, M.; Roshani, G. H.; Abdi, H.; Rezaei, A.; Mahtab, M.

2013-08-01

173

A Neural Network Approach for Inverse Kinematic of a SCARA Manipulator

Directory of Open Access Journals (Sweden)

Full Text Available Inverse kinematic is one of the most interesting problems of industrial robot. The inverse kinematics problem in robotics is about the determination of joint angles for a desired Cartesian position of the end effector. It comprises of the computation need to find the joint angles for a given Cartesian position and orientation of the end effectors to control a robot arm. There is no unique solution for the inverse kinematics thus necessitating application of appropriate predictive models from the soft computing domain. Artificial neural network is one such technique which can be gainfully used to yield the acceptable results. This paper proposes a structured artificial neural network (ANN model to find the inverse kinematics solution of a 4-dof SCARA manipulator. The ANN model used is a multi-layered perceptron neural network (MLPNN, wherein gradient descent type of learning rules is applied. An attempt has been made to find the best ANN configuration for the problem. It is found that multi-layered perceptron neural network gives minimum mean square error.

Panchanand Jha

2013-09-01

174

Neural-network classifiers for recognizing totally unconstrained handwritten numerals.

Artificial neural networks have been recognized as a powerful tool for pattern classification problems, but a number of researchers have also suggested that straightforward neural-network approaches to pattern recognition are largely inadequate for difficult problems such as handwritten numeral recognition. In this paper, we present three sophisticated neural-network classifiers to solve complex pattern recognition problems: multiple multilayer perceptron (MLP) classifier, hidden Markov model (HMM)/MLP hybrid classifier, and structure-adaptive self-organizing map (SOM) classifier. In order to verify the superiority of the proposed classifiers, experiments were performed with the unconstrained handwritten numeral database of Concordia University, Montreal, Canada. The three methods have produced 97.35%, 96.55%, and 96.05% of the recognition rates, respectively, which are better than those of several previous methods reported in the literature on the same database. PMID:18255609

Cho, S B

1997-01-01

175

Automatic Target Classification in SAR Images by Multilayer Back Propagation Neural Network

Digital Repository Infrastructure Vision for European Research (DRIVER)

In this study, a novel descriptive feature extraction method of Discrete Fourier transform and neural network classifier for classification of Synthetic Aperture Radar (SAR) images is proposed. The classification process has the following stages (1) Image Segmentation using statistical Region Merging (SRM) (2) Polar transform and Feature extraction using Discrete Fourier Transform (3) Neural Network classification using back propagation. The algorithm has been applied for the three classes of...

Vasuki, P.; Mohamed, S.; Mansoor Roomi

2012-01-01

176

Neural networks: a biased overview

International Nuclear Information System (INIS)

An overview of recent activity in the field of neural networks is presented. The long-range aim of this research is to understand how the brain works. First some of the problems are stated and terminology defined; then an attempt is made to explain why physicists are drawn to the field, and their main potential contribution. In particular, in recent years some interesting models have been introduced by physicists. A small subset of these models is described, with particular emphasis on those that are analytically soluble. Finally a brief review of the history and recent developments of single- and multilayer perceptrons is given, bringing the situation up to date regarding the central immediate problem of the field: search for a learning algorithm that has an associated convergence theorem

177

Two algorithms for neural-network design and training with application to channel equalization.

We describe two algorithms for designing and training neural-network classifiers. The first, the linear programming slab algorithm (LPSA), is motivated by the problem of reconstructing digital signals corrupted by passage through a dispersive channel and by additive noise. It constructs a multilayer perceptron (MLP) to separate two disjoint sets by using linear programming methods to identify network parameters. The second, the perceptron learning slab algorithm (PLSA), avoids the computational costs of linear programming by using an error-correction approach to identify parameters. Both algorithms operate in highly constrained parameter spaces and are able to exploit symmetry in the classification problem. Using these algorithms, we develop a number of procedures for the adaptive equalization of a complex linear 4-quadrature amplitude modulation (QAM) channel, and compare their performance in a simulation study. Results are given for both stationary and time-varying channels, the latter based on the COST 207 GSM propagation model. PMID:18252477

Sweatman, C Z; Mulgrew, B; Gibson, G J

1998-01-01

178

The key factors affecting the success of neural engineering using neural stem/progenitor cells (NSPCs) are the neuron quantity, the guidance of neurite outgrowth, and the induction of neurons to form functional synapses at synaptic junctions. Herein, a biomimetic material comprising a supported lipid bilayer (SLB) with adsorbed sequential polyelectrolyte multilayer (PEM) films was fabricated to induce NSPCs to form functional neurons without the need for serum and growth factors in a short-term culture. SLBs are suitable artificial substrates for neural engineering due to their structural similarity to synaptic membranes. In addition, PEM film adsorption provides protection for the SLB as well as the ability to vary the surface properties to evaluate the effects of physical and mechanical signals on NSPC differentiation. Our results revealed that NSPCs were inducible on SLB-PEM films consisting of up to eight alternating layers. In addition, the process outgrowth length, the percentage of differentiated neurons, and the synaptic function were regulated by the number of layers and the surface charge of the outermost layer. The average process outgrowth length was greater than 500 ?m on SLB-PLL/PLGA (n = 7.5) after only 3 days of culture. Moreover, the quantity and quality of the differentiated neurons were obviously enhanced on the SLB-PEM system compared with those on the PEM-only substrates. These results suggest that the PEM films can induce NSPC adhesion and differentiation and that an SLB base may enhance neuron differentiation and trigger the formation of functional synapses. PMID:25111699

Lee, I-Chi; Wu, Yu-Chieh

2014-08-27

179

Neural networks in front-end processing and control

International Nuclear Information System (INIS)

Research into neural networks has gained a large following in recent years. In spite of the long term timescale of this Artificial Intelligence research, the tools which the community is developing can already find useful applications to real practical problems in experimental research. One of the main advantages of the parallel algorithms being developed in AI is the structural simplicity of the required hardware implementation, and the simple nature of the calculations involved. This makes these techniques ideal for problems in which both speed and data volume reduction are important, the case for most front-end processing tasks. In this paper we illustrate the use of a particular neural network known as the Multi-Layer Perceptron as a method for solving several different tasks, all drawn from the field of Tokamak research. We also briefly discuss the use of the Multi-Layer Perceptron as a non-linear controller in a feedback loop. We outline the type of problem which can be usefully addressed by these techniques, even before the large-scale parallel processing hardware currently under development becomes cheaply available. We also present some of the difficulties encountered in applying these networks. (author) 13 figs., 9 refs

180

Automatic Target Classification in SAR Images by Multilayer Back Propagation Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available In this study, a novel descriptive feature extraction method of Discrete Fourier transform and neural network classifier for classification of Synthetic Aperture Radar (SAR images is proposed. The classification process has the following stages (1 Image Segmentation using statistical Region Merging (SRM (2 Polar transform and Feature extraction using Discrete Fourier Transform (3 Neural Network classification using back propagation. The algorithm has been applied for the three classes of military manmade objects (metal objects in SAR imagery is using MSTAR public release database. Experimental results are presente.

P. Vasuki

2012-12-01

181

Directory of Open Access Journals (Sweden)

Full Text Available Stock market prediction is one of the most important interesting areas of research in business. Stock markets prediction is normally assumed as tedious task since there are many factors influencing the market. The primary objective of this paper is to forecast trend closing price movement of Tehran Stock Exchange (TSE using financial accounting ratios from year 2003 to year 2008. The proposed study of this paper uses two approaches namely Artificial Neural Networks and multi-layer perceptron. Independent variables are accounting ratios and dependent variable of stock price , so the latter was gathered for the industry of Motor Vehicles and Auto Parts. The results of this study show that neural networks models are useful tools in forecasting stock price movements in emerging markets but multi-layer perception provides better results in term of lowering error terms.

Hossein Naderi

2012-08-01

182

A Study on Modeling of MIMO Channel by Using Different Neural Network Structures

Directory of Open Access Journals (Sweden)

Full Text Available Recognition of Radio Channel (channelParameters is one of Main Challenges in SignalTransformation, and has important role in cognitive radioapproach. Goal of this paper is “Channel modeling” to estimatecoefficients of transmission functions affected on data beingtransformed in the channel. We use Multilayer perceptron(MLPNeural Network with Back-propagation learning algorithm,block-structured Neural Network with Least Squares(LSmethod(cost function and a multilayer neural network withmultiple back-propagation(MBP learning algorithm for errorestimation. These networks will be trained with received signalsto be compatible with channel, then give us an estimation of thesecoefficients. Simulation will show that this MBP method is betterthan the other two method in error estimation. It has goodperformance and also consume less execution time. Then, we willuse this network for estimating coefficients of non-lineartransmission functions of actual radio channel.

N. Prabhakar

2012-11-01

183

Fast non-linear extraction of plasma equilibrium parameters using a neural network mapping

International Nuclear Information System (INIS)

The shaping of non-circular plasmas requires a non-linear mapping between the measured diagnostic signals and selected equilibrium parameters. The particular configuration of Neural Network known as the multi-layer perceptron provides a powerful and general technique for formulating an arbitrary continuous non-linear multi-dimensional mapping. This technique has been successfully applied to the extraction of equilibrium parameters from measurements of single-null diverted plasmas in the DIII-D tokamak; the results are compared with a purely linear mapping. The method is promising, and hardware implementation is straightforward. (author) 15 refs., 7 figs

184

Use of artificial neural networks for prognosis of charcoal prices in Minas Gerais

Digital Repository Infrastructure Vision for European Research (DRIVER)

Energy is an important factor of economic growth and is critical to the stability of a nation. Charcoal is a renewable energy resource and is a fundamental input to the development of the Brazilian forest-based industry. The objective of this study is to provide a prognosis of the charcoal price series for the year 2007 by using Artificial Neural Networks. A feedforward multilayer perceptron ANN was used, the results of which are close to reality. The main findings are that: real prices of ch...

Luiz Moreira Coelho Junior; José Luiz Pereira de Rezende; André Luiz França Batista; Adriano Ribeiro de Mendonça; Wilian Soares Lacerda

2013-01-01

185

Artificial neural network-derived trends in daily maximum surface ozone concentrations.

Interannual variability in meteorological conditions can confound attempts to identify changes in ozone concentrations driven by reduced precursor emissions. In this paper, a technique is described that attempts to maximize the removal of meteorological variability from a daily maximum ozone time series, thereby revealing longer term changes in ozone concentrations with increased confidence. The technique employs artificial neural network [multilayer perceptron (MLP)] models, and is shown to remove more of the meteorological variability from U.S. ozone data than does a Kolmogorov-Zurbenko (KZ) filter and conventional regression-based technique. PMID:11518294

Gardner, M; Dorling, S

2001-08-01

186

Solar Radiation Forecasting Using Ad-Hoc Time Series Preprocessing and Neural Networks.

Digital Repository Infrastructure Vision for European Research (DRIVER)

In this paper, we present an application of neural networks in the renewable energy domain. We have developed a methodology for the daily prediction of global solar radiation on a horizontal surface. We use an ad-hoc time series preprocessing and a Multi-Layer Perceptron (MLP) in order to predict solar radiation at daily horizon. First results are promising with nRMSE < 21% and RMSE < 998 Wh/m2. Our optimized MLP presents prediction similar to or even better than conventiona...

Paoli, Christophe; Voyant, Cyril; Muselli, Marc; Nivet, Marie Laure

2009-01-01

187

Alternative sensor system and MLP neural network for vehicle pedal activity estimation.

It is accepted that the activity of the vehicle pedals (i.e., throttle, brake, clutch) reflects the driver's behavior, which is at least partially related to the fuel consumption and vehicle pollutant emissions. This paper presents a solution to estimate the driver activity regardless of the type, model, and year of fabrication of the vehicle. The solution is based on an alternative sensor system (regime engine, vehicle speed, frontal inclination and linear acceleration) that reflects the activity of the pedals in an indirect way, to estimate that activity by means of a multilayer perceptron neural network with a single hidden layer. PMID:22319326

Wefky, Ahmed M; Espinosa, Felipe; Jiménez, José A; Santiso, Enrique; Rodríguez, José M; Fernández, Alfredo J

2010-01-01

188

Solar radiation forecasting using ad-hoc time series preprocessing and neural networks

In this paper, we present an application of neural networks in the renewable energy domain. We have developed a methodology for the daily prediction of global solar radiation on a horizontal surface. We use an ad-hoc time series preprocessing and a Multi-Layer Perceptron (MLP) in order to predict solar radiation at daily horizon. First results are promising with nRMSE < 21% and RMSE < 998 Wh/m2. Our optimized MLP presents prediction similar to or even better than conventional methods such as ARIMA techniques, Bayesian inference, Markov chains and k-Nearest-Neighbors approximators. Moreover we found that our data preprocessing approach can reduce significantly forecasting errors.

Paoli, Christophe; Muselli, Marc; Nivet, Marie-Laure

2009-01-01

189

How deals with discrete data for the reduction of simulation models using neural network

Simulation is useful for the evaluation of a Master Production/distribution Schedule (MPS). Also, the goal of this paper is the study of the design of a simulation model by reducing its complexity. According to theory of constraints, we want to build reduced models composed exclusively by bottlenecks and a neural network. Particularly a multilayer perceptron, is used. The structure of the network is determined by using a pruning procedure. This work focuses on the impact of discrete data on the results and compares different approaches to deal with these data. This approach is applied to sawmill internal supply chain

Thomas, Philippe

2009-01-01

190

Tissue-compliant neural implants from microfabricated carbon nanotube multilayer composite.

Current neural prosthetic devices (NPDs) induce chronic inflammation due to complex mechanical and biological reactions related, in part, to staggering discrepancies of mechanical properties with neural tissue. Relatively large size of the implants and traumas to blood-brain barrier contribute to inflammation reactions, as well. Mitigation of these problems and the realization of long-term brain interface require a new generation of NPDs fabricated from flexible materials compliant with the brain tissue. However, such materials will need to display hard-to-combine mechanical and electrical properties which are not available in the toolbox of classical neurotechnology. Moreover, these new materials will concomitantly demand different methods of (a) device micromanufacturing and (b) surgical implantation in brains because currently used processes take advantage of high stiffness of the devices. Carbon nanotubes (CNTs) serve as a promising foundation for such materials because of their record mechanical and electrical properties, but CNT-based tissue-compliant devices have not been realized yet. In this study, we formalize the mechanical requirements to tissue-compliant implants based on critical rupture strength of brain tissue and demonstrate that miniature CNT-based devices can satisfy these requirements. We fabricated them using MEMS-like technology and miniaturized them so that at least two dimensions of the electrodes would be comparable to brain tissue cells. The nanocomposite-based flexible neural electrodes were implanted into the rat motor cortex using a surgical procedure specifically designed for soft tissue-compliant implants. The post-surgery implant localization in the motor cortex was successfully visualized with magnetic resonance and photoacoustic imaging. In vivo functionality was demonstrated by successful registration of the low-frequency neural recording in the live brain of anesthetized rats. Investigation of inflammation processes around these electrodes will be required to establish their prospects as long-term neural electrodes. PMID:23930825

Zhang, Huanan; Patel, Paras R; Xie, Zhixing; Swanson, Scott D; Wang, Xueding; Kotov, Nicholas A

2013-09-24

191

Digital Repository Infrastructure Vision for European Research (DRIVER)

The multi-layer feed-forward neural network (MLFFNN) is applied in the context of river flow forecast combination, where a number of rainfall-runoff models are used simultaneously to produce an overall combined river flow forecast. The operation of the MLFFNN depends on the neuron transfer function, which is non-linear. These models, each having a different structure to simulate the perceived mechanisms of the runoff process, utilise the information carrying capacity of the model calibration ...

Shamseldin, Asaad Y.; Nasr, Ahmed Elssidig; O Connor, Kieran M.

2002-01-01

192

Simulation is often used to evaluate the relevance of a Directing Program of Production (PDP) or to evaluate its impact on detailed sc\\'enarii of scheduling. Within this framework, we propose to reduce the complexity of a model of simulation by exploiting a multilayer perceptron. A main phase of the modeling of one system using a multilayer perceptron remains the determination of the structure of the network. We propose to compare and use various pruning algorithms in order to determine the optimal structure of the network used to reduce the complexity of the model of simulation of our case of application: a sawmill.

Thomas, Philippe

2008-01-01

193

Training a multilayer neural network for the Euro-dollar (EUR/ USD exchange rate

Directory of Open Access Journals (Sweden)

Full Text Available A mathematical tool or model for predicting how an economic variable like the exchange rate (relative price between two currencies will respond is a very important need for investors and policy-makers. Most current techniques are based on statistics, particularly linear time series theory. Artificial neural networks (ANNs are mathematical models which try to emulate biological neural networks’ parallelism and nonlinearity; these models have been successfully applied in Economics and Engineering since the 1980s. ANNs appear to be an alternative for modelling the behaviour of financial variables which resemble (as first approximation a random walk. This paper reports the results of using ANNs for Euro/USD exchange rate trading and the usefulness of the algorithm for chemotaxis leading to training networks thereby maximising an objective function re predicting a trader’s profits. JEL: F310, C450.

Jaime Alberto Villamil Torres

2010-04-01

194

Time Series Data Mining in Rainfall Forecasting Using Artificial Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available Rainfall is very important parameter in hydrological model. Many techniques and models have been developed for rainfall time series prediction. In this study an artificial neural network (ANN based model was developed for rainfall time series forecasting. Proposed model used Multilayer perceptron (MLP network with back propagation algorithm for training. Discharge and rainfall data are took as input parameter for ANN model to predict rainfall time series. Data preprocessing and model’s sensitivity analysis were executed. Collected data is divided in three sets for optimal neural network training. The first set is the training set, used for calculate the gradient and updating the network weights and biases. The second set is the validation set. The error on the validation set is follow during the training process. The third set is test set. It is used to compare different models. Different topologies of Neural Networks were created with change in hidden layer, number of processing element and activation function. (MAE, Mean Squared error (MSE and correlation coefficient (CC are used to evaluate the model performance. On the basis of these evaluation parameter results, it is found that multilayer perceptron (MLP network predict more accurate than other traditional models.

Prince Gupta, S.K.Pandey

2014-01-01

195

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: English Abstract in portuguese Vinte variedades de soja (Glycine max), quatorze convencionais e seis variedades transgênicas (RR) foram analisadas quanto ao teor de proteína, ácido fítico, teor de óleo, fitosteróis, cinzas, minerais e ácidos graxos que foram tabelados e apresentados à rede neural do tipo perceptron de múltiplas c [...] amadas para a classificação e identificação quanto a região de plantio e quanto a variedade convencional ou transgênica. A rede neural utilizada classificou e testou corretamente 100% das amostras cultivadas por região. Para o banco de dados contendo informações sobre sojas transgênicas e convencionais foi obtido um desempenho de 94,43% no treinamento da rede, 83,30% no teste e 100% na validação. Abstract in english Twenty soybean (Glycine max) varieties, 14 conventional and 6 transgenic varieties were analyzed for protein content, phytic acid, oil content, phytosterols, ash, minerals and fatty acids. The data were tabled and presented to the multilayer perceptron neural network for classification and identific [...] ation of their planting region and whether they were a conventional or transgenic. The neural network used correctly classified and tested 100% of the samples cultivated per region. For the data bank containing information on transgenic and conventional soybean, a performance of 94.43% was obtained in the training of the neural network, 83.30% in the test and 100% in the validation.

Olívio F., Galão; Dionísio, Borsato; Jurandir P., Pinto; Jesuí V., Visentainer; Mercedes Concórdia, Carrão-Panizzi.

2011-01-01

196

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: English Abstract in portuguese Vinte variedades de soja (Glycine max), quatorze convencionais e seis variedades transgênicas (RR) foram analisadas quanto ao teor de proteína, ácido fítico, teor de óleo, fitosteróis, cinzas, minerais e ácidos graxos que foram tabelados e apresentados à rede neural do tipo perceptron de múltiplas c [...] amadas para a classificação e identificação quanto a região de plantio e quanto a variedade convencional ou transgênica. A rede neural utilizada classificou e testou corretamente 100% das amostras cultivadas por região. Para o banco de dados contendo informações sobre sojas transgênicas e convencionais foi obtido um desempenho de 94,43% no treinamento da rede, 83,30% no teste e 100% na validação. Abstract in english Twenty soybean (Glycine max) varieties, 14 conventional and 6 transgenic varieties were analyzed for protein content, phytic acid, oil content, phytosterols, ash, minerals and fatty acids. The data were tabled and presented to the multilayer perceptron neural network for classification and identific [...] ation of their planting region and whether they were a conventional or transgenic. The neural network used correctly classified and tested 100% of the samples cultivated per region. For the data bank containing information on transgenic and conventional soybean, a performance of 94.43% was obtained in the training of the neural network, 83.30% in the test and 100% in the validation.

Olívio F., Galão; Dionísio, Borsato; Jurandir P., Pinto; Jesuí V., Visentainer; Mercedes Concórdia, Carrão-Panizzi.

197

Directory of Open Access Journals (Sweden)

Full Text Available The main difficulty for natural disaster insurance derives from the uncertainty of an event's damages. Insurers cannot precisely appreciate the weight of natural hazards because of risk dependences. Insurability under uncertainty first requires an accurate assessment of entire damages. Insured and insurers both win when premiums calculate risk properly. In such cases, coverage will be available and affordable. Using the artificial neural network - a technique rooted in artificial intelligence - insurers can predict annual natural disaster losses. There are many types of artificial neural network models. In this paper we use the multilayer perceptron neural network, the most accommodated to the prediction task. In fact, if we provide the natural disaster explanatory variables to the developed neural network, it calculates perfectly the potential annual losses for the studied country.

Jemli Rim

2010-01-01

198

Practical Application of Neural Networks in State Space Control

DEFF Research Database (Denmark)

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 applied to a realistic process. The thesis therefore strives to provide a thorough treatment of two classes of neural network-based controllers, and to make a rigorous comparison between them and a classical linear controller. Thus, the thesis starts out with a short review of some relevant system 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 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 up for application of linear theory at each sample instant. The case study addressed in this work is an attemporator for a high-temperature steam circuit situated in a Danish powerplant, I/S Vestkraft unit 3. The attemporator is fitted with a nonlinear and nonconstant valve, so nonlinear and adaptive control is desired to control the steam temperature tightly. A second-order nonlinear model of the attemporator based on system identification with a multilayer perceptron network is found from data collected from the actual process. It is shown to be a highly satisfying prediction and simulation model of the process. With this model in place, we turn to the control concepts. A pole placement controller based on the sample-by-sample linearizations extracted from a multilayer perceptron state observer is first derived, and it is shown how to make the control concept adaptive by continuing the training online. Then the controller is shown to work on a simulation example. We also address the potential problem of too rapidly fluctuating parameters by including regularization in the learning rule. Next we develop a direct adaptive certainty-equivalence controller based on neurofuzzy models. The control loop is proven to be stable under certain assumptions, and we address the question of how many basis functions are necessary. It is shown that basis functions with compact supports, whose supports are not entered by a system trajectory, do not need parameter updates. Therefore, a system with bounded trajectories can be controlled by a finite-dimensional model. We also introduce a modification to the algorithm which-if an upper bound on the nonlinearity growth is known-enables us to remove a sector-boundedness assumption on the nonlinearity. Finally the control concepts are applied to the nonlinear simulation model discussed above, and it is seen that the neural network -based control concepts outperform a classical linear controller.

Bendtsen, Jan Dimon

1999-01-01

199

Directory of Open Access Journals (Sweden)

Full Text Available Recently, Artificial Neural Networks (ANN, which is mathematical modelingtools inspired by the properties of the biological neural system, has been typically used inthe studies of hydrological time series modeling. These modeling studies generally includethe standart feed forward backpropagation (FFBP algorithms such as gradient-descent,gradient-descent with momentum rate and, conjugate gradient etc. As the standart FFBPalgorithms have some disadvantages relating to the time requirement and slowconvergency in training, Newton and Levenberg-Marquardt algorithms, which arealternative approaches to standart FFBP algorithms, were improved and also used in theapplications. In this study, an application of Levenberg-Marquardt algorithm based ANN(LM-ANN for the modeling of monthly inflows of Demirkopru Dam, which is located inthe Gediz basin, was presented. The LM-ANN results were also compared with gradientdescentwith momentum rate algorithm based FFBP model (GDM-ANN. When thestatistics of the long-term and also seasonal-term outputs are compared, it can be seen thatthe LM-ANN model that has been developed, is more sensitive for prediction of theinflows. In addition, LM-ANN approach can be used for modeling of other hydrologicalcomponents in terms of a rapid assessment and its robustness.

Umut Okkan

2011-07-01

200

The margitron: a generalized perceptron with margin.

We identify the classical perceptron algorithm with margin as a member of a broader family of large margin classifiers, which we collectively call the margitron. The margitron, (despite its) sharing the same update rule with the perceptron, is shown in an incremental setting to converge in a finite number of updates to solutions possessing any desirable fraction of the maximum margin. We also report on experiments comparing the margitron with decomposition support vector machines, cutting-plane algorithms, and gradient descent methods on hard margin tasks involving linear kernels which are equivalent to 2-norm soft margin. Our results suggest that the margitron is very competitive. PMID:21216709

Panagiotakopoulos, Constantinos; Tsampouka, Petroula

2011-03-01

201

Parallel strategy for optimal learning in perceptrons

International Nuclear Information System (INIS)

We developed a parallel strategy for learning optimally specific realizable rules by perceptrons, in an online learning scenario. Our result is a generalization of the Caticha-Kinouchi (CK) algorithm developed for learning a perceptron with a synaptic vector drawn from a uniform distribution over the N-dimensional sphere, so called the typical case. Our method outperforms the CK algorithm in almost all possible situations, failing only in a denumerable set of cases. The algorithm is optimal in the sense that it saturates Bayesian bounds when it succeeds.

202

Neural networks in structural analysis and design - An overview

The present paper provides an overview of the state-of-the-art in the application of neural networks in problems of structural analysis and design, including a survey of published applications in structural engineering. Such applications have included, among others, the use of neural networks in modeling nonlinear analysis of structures, as a rapid reanalysis capability in optimal design, and in developing problem parameter sensitivity of optimal solutions for use in multilevel decomposition based design. While most of the applications reported in the literature have been restricted to the use of the multilayer perceptron architecture and minor variations thereof, other network architectures have also been successfully explored, including the ART network, the counterpropagation network and the Hopfield-Tank model.

Hajela, P.; Berke, L.

1992-01-01

203

Directory of Open Access Journals (Sweden)

Full Text Available Interpolation of a spatially continuous variable from point samples is an important field in spatial analysis and surface models for geosciences. In this study, spatial interpolation methods which are Inverse Distance Weighted (IDW, Ordinary Kriging (OK, Modified Shepard's (MS, Multiquadric Radial Basis Function (MRBF and Triangulation with Linear (TWL, and Multi-Layer Perceptron (MLP which is an Artificial Neural Networks (ANN method were compared in order to predict height for different point distributions such as curvature, grid, random and uniform on a Digital Elevation Model which is an USGS National Elevation Dataset (NED. This study also aims to quantify the effects of topographic variability and sampling density. Errors of different interpolations and ANN prediction were evaluated for different point distributions and three different crosssections on the characteristic parts of the surface were selected and analyzed. Generally, OK, MS, MRBF and TWL gave promising results and were more effective in terms of characteristics of surface than MLP and IDW. Although MLP simplified the contours obtained from predicted heights, it was a satisfactory predictor for curvature, grid, random and uniform distributions ; Interpolacija prostorsko zvezne spremenljivke iz to?kovnih primerov je v geoznanosti pomembno podro?je prostorske analize in modelov površja. V opisani študiji je bila izvedena primerjava interpolacijskih metod v trirazsežnem prostoru, in sicer so to metoda z inverzno uteženo razdaljo (IDW, navadni kriging (OK, modificirana Shepardova metoda (MS, multikvadri?na radialna funkcija (MRBF in triangulacija z linearno interpolacijo (TWL ter ve?slojni perceptron (MLP, ki je predstavnik umetnih nevronskih mrež (ANN. Cilj je bil napovedati višino za razli?ne geometrijske razporeditve to?k, kot so ukrivljenost, mreža, naklju?na in enotna porazdelitev na digitalnem modelu višin, ki je podatkovni niz digitalnega modela višin ameriške geološke službe USGS. Namen študije je koli?insko opredeliti u?inek topografske variabilnosti in gostote vzor?enja. Napake razli?nih interpolacij in napovedi z umetnimi nevronskimi mrežami so bile ovrednotene glede na razli?ne geometrijske porazdelitve to?k, izbrani in analizirani so bili tri razli?ni prerezi zna?ilnih delov površja. Na splošno se je izkazalo, da metode navadni kriging (OK, modificirana Shepardova metoda (MS, multikvadri?na radialna funkcija (MRBF in triangulacija z linearno interpolacijo (TWL dajejo boljše rezultate ter so bolj u?inkovite glede zna?ilnosti površja kot ve?slojni perceptron (MLP in metoda z uteženo inverzno razdaljo (IDW. ?eprav je ve?slojni perceptron (MLP poenostavil obrise, pridobljene iz napovedanih višin, se je izkazal kot zadovoljiv pri napovedovanju ukrivljenosti ter dolo?itvi celi?ne mreže za naklju?ne in znane geometrijske porazdelitve to?k

Kutalmis Gumus

2013-01-01

204

CONTROL OF NONLINEAR PROCESS USING NEURAL NETWORK BASED MODEL PREDICTIVE CONTROL

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents a Neural Network based Model Predictive Control (NNMPC strategy to control nonlinear process. Multilayer Perceptron Neural Network (MLP is chosen to represent a Nonlinear Auto Regressive with eXogenous signal (NARX model of a nonlinear system. NARX dynamic model is based on feed-forward architecture and offers good approximation capabilities along with robustness and accuracy. Based on the identified neural model, a generalized predictive control (GPC algorithm is implemented to control the composition in acontinuous stirred tank reactor (CSTR, whose parameters are optimally determined by solving quadratic performance index using well known Levenberg-Marquardt and Quasi-Newton algorithm. NNMPC is tuned by selecting few horizon parameters and weighting factor. The tracking performance of the NNMPC is tested using different amplitude function as a reference signal on CSTR application. Also the robustness and performance is tested in the presence of disturbance on random reference signal.

Dr.A.TRIVEDI

2011-04-01

205

An Approach to Neural Network Based Pattern Classifier for Printed Bengali Characters

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, we have designed a Neural Network based pattern classifier for recognizing Bengali printed characters. Here view-based approach is used for extracting features from individual characters and a neural network based classifier is built to analyze the performance of the view-based approach in various experimental setups. Different Bengali character samples have been taken and whole image of individual character is considered for view based analysis. The characteristic points are extracted from the characters using left-right view based approach. These points are then used to form a feature vector which represents the given character. Multi-Layer Perceptrons Neural network has been used and it was trained by back propagation algorithm to create this recognition engine. Internal shape of each character has been considered to generate the feature vector for individual images.

sabyasachi samanta

2011-04-01

206

Inversion of rocket-borne photometer measurements by an artificial neural network technique

International Nuclear Information System (INIS)

Complete text of publication follows. The inverse problem to retrieve useful airglow volume emission rate profiles from rocket-borne photometer measurements has been solved by adopting the well-characterized spectral photometric methods. An alternative recovery method based on artificial neural network (ANN) is presented. In this work, a multilayer perceptron neural network was trained with a range of cases from the empirical and experimental volume emission rate profiles. A numerical experiment was also carried out with synthetic experimental data considering a noise level of 5%. Integrated emission profiles measured by a Brazilian sounding rocket experiment launched from an Equatorial station were taken as the input data. From the results obtained it may be concluded that the ANN technique is a convenient tool to recover volume emission rate profiles. The advantages of using neural network based systems are related to their intrinsic features of parallelism, after trained, the networks are much faster than traditional inversion approaches.

207

Effect of direction on wind speed estimation in complex terrain using neural networks

Energy Technology Data Exchange (ETDEWEB)

A method of estimating the annual average wind speed at a selected site using neural networks is presented. The method proposed uses only a few measurements taken at the selected site in a short time period and data collected at nearby fixed stations. The neural network used in this study is a multilayer perceptron with one hidden layer of 15 neurons, trained by the Bayesian regularization algorithm. The number of inputs that must be used in the neural network was analyzed in detail, and results suggest that only wind speed and direction data for a single station are required. In sites of complex terrain, direction is a very important input that can cause a decrease of 23% in root mean square (RMS). The results obtained by simulating the annual average wind speed at the selected site based on data from nearby stations are satisfactory, with errors below 2%. (author)

Lopez, P.; Velo, R.; Maseda, F. [Department of Agroforestry Engineering, Higher Polytechnic School. University of Santiago de Compostela, Campus Universitario, s/n, 27002, Lugo (Spain)

2008-10-15

208

Thrips (Thysanoptera) identification using artificial neural networks.

We studied the use of a supervised artificial neural network (ANN) model for semi-automated identification of 18 common European species of Thysanoptera from four genera: Aeolothrips Haliday (Aeolothripidae), Chirothrips Haliday, Dendrothrips Uzel, and Limothrips Haliday (all Thripidae). As input data, we entered 17 continuous morphometric and two qualitative two-state characters measured or determined on different parts of the thrips body (head, pronotum, forewing and ovipositor) and the sex. Our experimental data set included 498 thrips specimens. A relatively simple ANN architecture (multilayer perceptrons with a single hidden layer) enabled a 97% correct simultaneous identification of both males and females of all the 18 species in an independent test. This high reliability of classification is promising for a wider application of ANN in the practice of Thysanoptera identification. PMID:18423077

Fedor, P; Malenovský, I; Vanhara, J; Sierka, W; Havel, J

2008-10-01

209

Electroactive scaffolds that are passively conductive and able to transmit applied electrical stimuli are of increasing importance for neural tissue engineering. Here, we report a process of rendering both 2D and 3D polymer scaffolds electrically conducting, while also enhancing neuron attachment. Graphene-heparin/poly-l-lysine polyelectrolytes were assembled via layer-by-layer (LbL) deposition onto 2D surfaces and 3D electrospun nanofibers. The employed LbL coating technique in this work enables the electro- and biofunctionalization of complex 3D scaffold structures. LbL assembly was characterized by a steady mass increase during the in situ deposition process in 2D, with regular step changes in hydrophobicity. Uniform coverage of the graphene/polyelectrolyte coatings was also achieved on nanofibers, with hydrodynamic flow and post-thermal annealing playing an important role in controlling sheet resistance of 2D surfaces and nanofibers. Cell culture experiments showed that both 2D and 3D graphene-PEMs supported neuron cell adhesion and neurite outgrowth, with no appreciable cell death. This electroactive scaffold modification may therefore assist in neuronal regeneration, for creating functional and biocompatible polymer scaffolds for electrical entrainment or biosensing applications. PMID:22809168

Zhou, Kun; Thouas, George A; Bernard, Claude C; Nisbet, David R; Finkelstein, David I; Li, Dan; Forsythe, John S

2012-09-26

210

Directory of Open Access Journals (Sweden)

Full Text Available Introduction: We aimed to develop a classification method to discriminate ventricular septal defect and atrial septal defect by using severalhemodynamic parameters.Patients and Methods: Forty three patients (30 atrial septal defect, 13 ventricular septal defect; 26 female, 17 male with documentedhemodynamic parameters via cardiac catheterization are included to study. Such parameters as blood pressure values of different areas,gender, age and Qp/Qs ratios are used for classification. Parameters, we used in classification are determined by divergence analysismethod. Those parameters are; i pulmonary artery diastolic pressure, ii Qp/Qs ratio, iii right atrium pressure, iv age, v pulmonary arterysystolic pressure, vi left ventricular sistolic pressure, vii aorta mean pressure, viii left ventricular diastolic pressure, ix aorta diastolicpressure, x aorta systolic pressure. Those parameters detected from our study population, are uploaded to multi-layered artificial neuralnetwork and the network was trained by genetic algorithm.Results: Trained cluster consists of 14 factors (7 atrial septal defect and 7 ventricular septal defect. Overall success ratio is 79.2%, andwith a proper instruction of artificial neural network this ratio increases up to 89%.Conclusion: Parameters, belonging to artificial neural network, which are needed to be detected by the investigator in classical methods,can easily be detected with the help of genetic algorithms. During the instruction of artificial neural network by genetic algorithms, boththe topology of network and factors of network can be determined. During the test stage, elements, not included in instruction cluster, areassumed as in test cluster, and as a result of this study, we observed that multi-layered artificial neural network can be instructed properly,and neural network is a successful method for aimed classification.

Mustafa Y?ld?z

2012-08-01

211

Simulation Study of Mass Transfer Coefficient in Slurry Bubble Column Reactor Using Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available The objective of this study was to develop neural network algorithm, (Multilayer Perceptron, based correlations for the prediction overall volumetric mass-transfer coefficient (kLa, in slurry bubble column for gas-liquid-solid systems. The Multilayer Perceptron is a novel technique based on the feature generation approach using back propagation neural network. Measurements of overall volumetric mass transfer coefficient were made with the air - Water, air - Glycerin and air - Alcohol systems as the liquid phase in bubble column of 0.15 m diameter. For operation with gas velocity in the range 0-20 cm/sec, the overall volumetric mass transfer coefficient was found to decrease with increasing solid concentration. From the experimental work 1575 data points for three systems, were collected and used to predicate kLa. Using SPSS 17 software, predicting of overall volumetric mass-transfer coefficient (kLa was carried out and an output of 0.05264 sum of square error was obtained for trained data and 0.01064 for test data.

Safa A. Al-Naimi

2013-01-01

212

Financial Time Series Prediction Using Spiking Neural Networks

In this paper a novel application of a particular type of spiking neural network, a Polychronous Spiking Network, was used for financial time series prediction. It is argued that the inherent temporal capabilities of this type of network are suited to non-stationary data such as this. The performance of the spiking neural network was benchmarked against three systems: two “traditional”, rate-encoded, neural networks; a Multi-Layer Perceptron neural network and a Dynamic Ridge Polynomial neural network, and a standard Linear Predictor Coefficients model. For this comparison three non-stationary and noisy time series were used: IBM stock data; US/Euro exchange rate data, and the price of Brent crude oil. The experiments demonstrated favourable prediction results for the Spiking Neural Network in terms of Annualised Return and prediction error for 5-Step ahead predictions. These results were also supported by other relevant metrics such as Maximum Drawdown and Signal-To-Noise ratio. This work demonstrated the applicability of the Polychronous Spiking Network to financial data forecasting and this in turn indicates the potential of using such networks over traditional systems in difficult to manage non-stationary environments. PMID:25170618

Reid, David; Hussain, Abir Jaafar; Tawfik, Hissam

2014-01-01

213

Neural network classifier of attacks in IP telephony

Various types of monitoring mechanism allow us to detect and monitor behavior of attackers in VoIP networks. Analysis of detected malicious traffic is crucial for further investigation and hardening the network. This analysis is typically based on statistical methods and the article brings a solution based on neural network. The proposed algorithm is used as a classifier of attacks in a distributed monitoring network of independent honeypot probes. Information about attacks on these honeypots is collected on a centralized server and then classified. This classification is based on different mechanisms. One of them is based on the multilayer perceptron neural network. The article describes inner structure of used neural network and also information about implementation of this network. The learning set for this neural network is based on real attack data collected from IP telephony honeypot called Dionaea. We prepare the learning set from real attack data after collecting, cleaning and aggregation of this information. After proper learning is the neural network capable to classify 6 types of most commonly used VoIP attacks. Using neural network classifier brings more accurate attack classification in a distributed system of honeypots. With this approach is possible to detect malicious behavior in a different part of networks, which are logically or geographically divided and use the information from one network to harden security in other networks. Centralized server for distributed set of nodes serves not only as a collector and classifier of attack data, but also as a mechanism for generating a precaution steps against attacks.

Safarik, Jakub; Voznak, Miroslav; Mehic, Miralem; Partila, Pavol; Mikulec, Martin

2014-05-01

214

Nonlinear Process Identification using Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available In industry process control, the model identification of nonlinear systems are always difficult problems. The main aim of this paper is to establish a reliable model for the nonlinear process. In many applications, development of empirical nonlinear model from dynamic plant data. This process is known as ‘Nonlinear System Identification’. Artificial neural networks are the most popular frame-work for empirical model development .In order to obtain this reliable model for the process dynamics, the neural black-box identification by means of a Nonlinear Autoregressive exogenous input (NARMAX model has been chosen in this study. The model is implemented by training a Multi-Layer Perceptron Artificial Neural Network (MLP-ANN with input-output experimental data is found and results shown that the neural model successfully predicts the evolution of the product composition. The simulation result illustrates the validity and feasibility of the nonlinear model identification. Trained data obtained from nonlinear process identification, can be used to control the nonlinear system

Miss.Mali Priyadarshani S. *1,

2014-06-01

215

Learning action representations using kernel perceptrons

Digital Repository Infrastructure Vision for European Research (DRIVER)

Action representation is fundamental to many aspects of cognition, including language. Theories of situated cognition suggest that the form of such representation is distinctively determined by grounding in the real world. This thesis tackles the question of how to ground action representations, and proposes an approach for learning action models in noisy, partially observable domains, using deictic representations and kernel perceptrons. Agents operating in real-world setti...

Mourao, Kira Margaret Thom

2012-01-01

216

Data-driven automated acoustic analysis of human infant vocalizations using neural network tools.

Acoustic analysis of infant vocalizations has typically employed traditional acoustic measures drawn from adult speech acoustics, such as f(0), duration, formant frequencies, amplitude, and pitch perturbation. Here an alternative and complementary method is proposed in which data-derived spectrographic features are central. 1-s-long spectrograms of vocalizations produced by six infants recorded longitudinally between ages 3 and 11 months are analyzed using a neural network consisting of a self-organizing map and a single-layer perceptron. The self-organizing map acquires a set of holistic, data-derived spectrographic receptive fields. The single-layer perceptron receives self-organizing map activations as input and is trained to classify utterances into prelinguistic phonatory categories (squeal, vocant, or growl), identify the ages at which they were produced, and identify the individuals who produced them. Classification performance was significantly better than chance for all three classification tasks. Performance is compared to another popular architecture, the fully supervised multilayer perceptron. In addition, the network's weights and patterns of activation are explored from several angles, for example, through traditional acoustic measurements of the network's receptive fields. Results support the use of this and related tools for deriving holistic acoustic features directly from infant vocalization data and for the automatic classification of infant vocalizations. PMID:20370038

Warlaumont, Anne S; Oller, D Kimbrough; Buder, Eugene H; Dale, Rick; Kozma, Robert

2010-04-01

217

We propose an approach using artificial neural networks to classify masses in mammograms as malignant or benign. Single-layer and multilayer perceptron networks are used in a study on perceptron topologies and training procedures for pattern classification of breast masses. The contours of a set of 111 regions on mammograms related to breast masses and tumors are manually delineated and represented by polygonal models for shape analysis. Ribbons of pixels are extracted around the boundaries of a subset of 57 masses by dilating and eroding the contours. Three shape factors, three measures of edge sharpness, and 14 texture features based on gray-level co-occurrence matrices of the pixels in the ribbons are computed. Several combinations of the features are used with perceptrons of varying topology and training procedures for the classification of benign masses and malignant tumors. The results are compared in terms of the area Az under the receiver operating characteristics curve. Values of Az up to 0.99 are obtained with the shape factors and texture features. However, only feature sets that included at least one shape factor provide consistently high performance with respect to variations in network topology and training.

André, Túlio C. S. S.; Rangayyan, Rangaraj M.

2006-01-01

218

Ensemble learning of linear perceptron; Online learning theory

Within the framework of on-line learning, we study the generalization error of an ensemble learning machine learning from a linear teacher perceptron. The generalization error achieved by an ensemble of linear perceptrons having homogeneous or inhomogeneous initial weight vectors is precisely calculated at the thermodynamic limit of a large number of input elements and shows rich behavior. Our main findings are as follows. For learning with homogeneous initial weight vectors, the generalization error using an infinite number of linear student perceptrons is equal to only half that of a single linear perceptron, and converges with that of the infinite case with O(1/K) for a finite number of K linear perceptrons. For learning with inhomogeneous initial weight vectors, it is advantageous to use an approach of weighted averaging over the output of the linear perceptrons, and we show the conditions under which the optimal weights are constant during the learning process. The optimal weights depend on only correlat...

Hara, K; Hara, Kazuyuki; Okada, Masato

2004-01-01

219

Neural network based near- lossless compression of EEG signals with non uniform quantization.

Efficient compression technique is highly essential for the transmission and storage of large amount of biomedical signals. In this paper, a near- lossless scheme for the compression of EEG signals using artificial neural networks is proposed. The error (residue) signals which is obtained due to the difference between the original and the predicted EEG signals are thresolded based on a term referred as absolute error limit (AEL) such that, any error samples above the limit require more number of bits than the samples below the limit that require less number of bits. The thresholded error samples are quantized in a non-uniform manner by varying the actual bits assigned to the error samples. An arithmetic encoder is further used to improve the compression efficiency. Three adaptive neural network models, namely, single and multilayer perceptrons and Elman neural network and two classical adaptive predictors such as autoregressive model(AR) and normalized least mean-square FIR filter are used. EEG signals recorded under different physiological conditions are considered and the performance of the proposed scheme is evaluated in terms of compression ratio and the fidelity parameter, percent of root-mean-square-difference (PRD). It is found from the experimental results that the variation of error limit and quantization step decides the overall compression performance. Single- layer perceptron yields the best compression results in terms of utilizing less bit rate as well achieving low PRD values compared to other predictors. PMID:18002685

Sriraam, N

2007-01-01

220

Directory of Open Access Journals (Sweden)

Full Text Available The multilayer perceptron network was used to classify the gasoline. The main parameters used in the classification were established by the Ordinance nº 309 of the Agência Nacional do Petróleo, but without informing the network the legal limits of these parameters. The network used had 10 neurons in a single hidden layer, learning rate of 0.04 and 250 training epochs. The application of artificial neural network served classify 100% of the commercialized gas in the region of Londrina-PR and to identify the tampered gasoline even those suspected of tampering.

Dionísio Borsato

2009-01-01

221

Radar Signal Detection In Non-Gaussian Noise Using RBF Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, we suggest a neural network signal detector using radial basis function (RBF network. We employ this RBF Neural detector to detect the presence or absence of a known signal corrupted by different Gaussian, non-Gaussian and impulsive noise components. In case of non-Gaussian noise, experimental results show that RBF network signal detector has significant improvement in performance characteristics. Detection capability is better than to those obtained with multilayer perceptrons (BP and optimum matched filter (MF detector. This signal detector is also tested on the simulated signals impacted by impulsive noise produced by atmospheric events and short lived echoes from meteor trains. Tested Results show, improved detection capability to impulsive noise compare to BP signal detector. It also show better performance as a function of signal-tonoise ratio compared to BP and MF.

D. G. Khairnar

2008-01-01

222

Foreground removal from Planck Sky Model temperature maps using a MLP neural network

DEFF Research Database (Denmark)

Unfortunately, the Cosmic Microwave Background (CMB) radiation is contaminated by emission originating in the Milky Way (synchrotron, free-free and dust emission). Since the cosmological information is statistically in nature, it is essential to remove this foreground emission and leave the CMB with no systematic errors. To demonstrate the feasibility of a simple multilayer perceptron (MLP) neural network for extracting the CMB temperature signal, we have analyzed a specific data set, namely the Planck Sky Model maps, developed for evaluation of different component separation methods before including them in the Planck data analysis pipeline. It is found that a MLP neural network can provide a CMB map of about 80% of the sky to a very high degree uncorrelated with the foreground components. Also the derived power spectrum shows little evidence for systematic errors.

NØrgaard-Nielsen, Hans Ulrik; Hebert, K.

2009-01-01

223

Prediction of Force Measurements of a Microbend Sensor Based on an Artificial Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available Artificial neural network (ANN based prediction of the response of a microbend fiber optic sensor is presented. To the best of our knowledge no similar work has been previously reported in the literature. Parallel corrugated plates with three deformation cycles, 6 mm thickness of the spacer material and 16 mm mechanical periodicity between deformations were used in the microbend sensor. Multilayer Perceptron (MLP with different training algorithms, Radial Basis Function (RBF network and General Regression Neural Network (GRNN are used as ANN models in this work. All of these models can predict the sensor responses with considerable errors. RBF has the best performance with the smallest mean square error (MSE values of training and test results. Among the MLP algorithms and GRNN the Levenberg-Marquardt algorithm has good results. These models successfully predict the sensor responses, hence ANNs can be used as useful tool in the design of more robust fiber optic sensors.

Kemal Fidanboylu

2009-09-01

224

A hybrid Evolutionary Functional Link Artificial Neural Network for Data mining and Classification

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents a specific structure of neural network as the functional link artificial neural network (FLANN. This technique has been employed for classification tasks of data mining. In fact, there are a few studies that used this tool for solving classification problems. In this present research, we propose a hybrid FLANN (HFLANN model, where the optimization process is performed using 3 known population based techniques such as genetic algorithms, particle swarm and differential evolution. This model will be empirically compared to FLANN based back-propagation algorithm and to others classifiers as decision tree, multilayer perceptron based back-propagation algorithm, radical basic function, support vector machine, and K-nearest Neighbor. Our results proved that the proposed model outperforms the other single model. (Abstract

Faissal MILI

2012-08-01

225

Directory of Open Access Journals (Sweden)

Full Text Available Thirteen cytology of fine needle aspiration image (i.e. cellularity, background information, cohesiveness, significant stromal component, clump thickness, nuclear membrane, bare nuclei, normal nuclei, mitosis, nucleus stain, uniformity of cell, fragility and number of cells in cluster are evaluated their possibility to be used as input data for artificial neural network in order to classify the breast pre-cancerous cases into four stages, namely malignant, fibroadenoma, fibrocystic disease, and other benign diseases. A total of 1300 reported breast pre-cancerous cases which was collected from Penang General Hospital and Hospital Universiti Sains Malaysia, Kelantan, Malaysia was used to train and test the artificial neural networks. The diagnosis system which was developed using the Hybrid Multilayered Perceptron and trained using Modified Recursive Prediction Error produced excellent diagnosis performance with 100% accuracy, 100% sensitivity and 100% specificity.

Nor A.M. Isa

2007-01-01

226

Artificial neural network for modeling the extraction of aromatic hydrocarbons from lube oil cuts

Energy Technology Data Exchange (ETDEWEB)

An artificial neural network (ANN) approach was used to obtain a simulation model to predict the rotating disc contactor (RDC) performance during the extraction of aromatic hydrocarbons from lube oil cuts, to produce a lubricating base oil using furfural as solvent. The field data used for training the ANN model was obtained from a lubricating oil production company. The input parameters of the ANN model were the volumetric flow rates of feed and solvent, the temperatures of feed and solvent, and the disc rotation rate. The output parameters were the volumetric flow rate of the raffinate phase and the extraction yield. In this study, a feed-forward multi-layer perceptron neural network was successfully used to demonstrate the complex relationship between the mentioned input and output parameters. (Copyright copyright 2011 WILEY-VCH Verlag GmbH and Co. KGaA, Weinheim)

Mehrkesh, A.H.; Hajimirzaee, S. [Islamic Azad University, Majlesi Branch, Isfahan (Iran, Islamic Republic of); Hatamipour, M.S.; Tavakoli, T. [Department of Chemical Engineering, University of Isfahan, Isfahan (Iran, Islamic Republic of)

2011-03-15

227

DEFF Research Database (Denmark)

The purpose of this study was to create an empirical model for assessing the landslide risk potential at Savadkouh Azad University, which is located in the rural surroundings of Savadkouh, about 5 km from the city of Pol-Sefid in northern Iran. The soil longitudinal profile of the city of Babol, located 25 km from the Caspian Sea, also was predicted with an artificial neural network (ANN). A multilayer perceptron neural network model was applied to the landslide area and was used to analyze specific elements in the study area that contributed to previous landsliding events. The ANN models were trained with geotechnical data obtained from an investigation of the study area. The quality of the modeling was improved further by the application of some controlling techniques involved in ANN. The observed >90% overall accuracy produced by the ANN technique in both cases is promising for future studies in landslide susceptibility zonation.

Farrokhzad, F.; Barari, Amin

2011-01-01

228

A Comparison between Neural Networks and Wavelet Networks in Nonlinear System Identification

Directory of Open Access Journals (Sweden)

Full Text Available In this study, identification of a nonlinear function will be presented by neural network and wavelet network methods. Behavior of a nonlinear system can be identified by intelligent methods. Two groups of the most common and at the same time the most effective of neural networks methods are multilayer perceptron and radial basis function that will be used for nonlinear system identification. The selected structure is series - parallel method that after network training by a series of training random data, the output is estimated and the nonlinear function is compared to a sinusoidal input. Then, wavelet network is used for identification and we will use Orthogonal Least Squares (OLS method for wavelet selection to reduce the volume of calculations and increase the convergence speed.

S. Ehsan Razavi

2012-01-01

229

Foreground removal from CMB temperature maps using an MLP neural network

One of the main obstacles in extracting the Cosmic Microwave Background (CMB) signal from observations in the mm-submm range is the foreground contamination by emission from galactic components: mainly synchrotron, free-free and thermal dust emission. Due to the statistical nature of the intrinsic CMB signal it is essential to minimize the systematic errors in the CMB temperature determinations. Following the available knowledge of the spectral behavior of the galactic foregrounds simple, power law-like spectra have been assumed. The feasibility of using a simple neural network for extracting the CMB temperature signal from the combined CMB and foreground signals has been investigated. As a specific example, we have analysed simulated data, like that expected from the ESA Planck Surveyor mission. A simple multilayer perceptron neural network with 2 hidden layers can provide temperature estimates, over more than 80 percent of the sky, that are to a high degree uncorrelated with the foreground signals. A single...

Norgaard-Nielsen, H U

2008-01-01

230

A Pareto evolutionary artificial neural network approach for remote sensing image classification

This paper presents a Pareto evolutionary artificial neural network (Pareto-EANN) approach based on the evolutionary algorithms for multiobjective optimization augmented with local search for the classification of remote sensing image. Its novelty lies in the use of a multiobjective genetic algorithm where single hidden layers Multilayer Perceptrons (MLP) are employed to indicate the accuracy/complexity trade-off. Some advantages of this approach include the ability to accommodate multiple criteria such as accuracy of the classifier and number of hidden units. We compared Pareto-EANN classifiers results of the classification of remote sensing image against standard backpropagation neural network classifiers and EANN classifiers; we show experimentally the efficiency of the proposed methodology.

Liu, Fujiang; Wu, Xincai; Guo, Yan; Sun, Huashan; Zhou, Feng; Mei, Linlu

2006-10-01

231

Sensor fusion for the navigation of an autonomous guided vehicle using neural networks

A sensor fusion method for navigation of an Autonomous Guided Vehicle robot using Artificial Neural Network is described. Robot navigation is defined as the guiding of a mobile robot to a desired destination or along a desired path in an environment characterized by a terrain and a set of distinct objects, such as obstacles and landmarks. The low-level sensor fusion technique is used for direct integration of sensor data, resulting in parameter and state estimates. The multi-layered perceptron, with a single hidden layer in neural network structure, and the back- propagation algorithm are employed for the mobile robot's navigation and for obstacle avoidance. The significance of this work lies in the development of a new estimation method for mobile robot obstacle avoidance and guidance.

Cao, Jin; Hall, Ernest L.

1998-10-01

232

We propose intelligent methods for classifying three different muscle types, i.e. biceps, frontallis and abductor pollicis brevis muscles, with low computational complexity. For this aim, electromyogram (EMG) signals are recorded and modelled by using an auto-regressive (AR) model. As the size of the EMG signals is usually large, the computational complexity of artificial neural network (ANN) systems drastically increases. Therefore, in the proposed scheme EMG signals are pre-processed by using a wavelet transform and then they are modelled by employing an AR approach. The AR coefficients are used to train and test the ANNs. Experimental results show that the highest achieved classification accuracy is more than 95% in the case of EMG signals pre-processed by wavelet transform. The wavelet transform-based pre-processing significantly increases the performance rates compared to standard multilayer perceptron and general regression neural networks algorithms. PMID:20645198

Ozsert, M; Yavuz, O; Durak-Ata, L

2011-06-01

233

Neural Network on Photodegradation of Octylphenol using Natural and Artificial UV Radiation

Directory of Open Access Journals (Sweden)

Full Text Available The present paper comes up with an experimental design meant to point out the factors interferingin octylphenol’s degradation in surface waters under solar radiation, underlining each factor’sinfluence on the process observable (concentration of p-octylphenol. Multiple linear regressionanalysis and artificial neural network (Multi-Layer Perceptron type were applied in order to obtaina mathematical model capable to explain the action of UV-light upon synthetic solutions of OP inultra-pure water (MilliQ type. Neural network proves to be the most efficient method in predictingthe evolution of OP concentration during photodegradation process. Thus, determination in neuralnetwork’s case has almost double value versus the regression analysis.

Lorentz JÄNTSCHI

2011-09-01

234

Neural network controller for Active Demand-Side Management with PV energy in the residential sector

International Nuclear Information System (INIS)

Highlights: ? We have developed a neural controller for Active Demand-Side Management. ? The controller consists of Multilayer Perceptrons evolved with a genetic algorithm. ? The architecture of the controller is distributed and modular. ? The simulations show that the electrical local behavior improves. ? Active Demand-Side Management helps users to control his energy behaviour. -- Abstract: In this paper, we describe the development of a control system for Demand-Side Management in the residential sector with Distributed Generation. The electrical system under study incorporates local PV energy generation, an electricity storage system, connection to the grid and a home automation system. The distributed control system is composed of two modules: a scheduler and a coordinator, both implemented with neural networks. The control system enhances the local energy performance, scheduling the tasks demanded by the user and maximizing the use of local generation.

235

Use of artificial neural networks to identify the origin of green macroalgae

This study demonstrates application of artificial neural networks (ANNs) for identifying the origin of green macroalgae ( Enteromorpha sp. and Cladophora sp.) according to their concentrations of Cd, Cu, Ni, Zn, Mn, Pb, Na, Ca, K and Mg. Earlier studies confirmed that algae can be used for biomonitoring surveys of metal contaminants in coastal areas of the Southern Baltic. The same data sets were classified with the use of different structures of radial basis function (RBF) and multilayer perceptron (MLP) networks. The selected networks were able to classify the samples according to their geographical origin, i.e. Southern Baltic, Gulf of Gda?sk and Vistula Lagoon. Additionally in the case of macroalgae from the Gulf of Gda?sk, the networks enabled the discrimination of samples according to areas of contrasting levels of pollution. Hence this study shows that artificial neural networks can be a valuable tool in biomonitoring studies.

?bikowski, Rados?aw

2011-08-01

236

Digital Repository Infrastructure Vision for European Research (DRIVER)

In the present research, possibility of predicting average summer-monsoon rainfall over India has been analyzed through Artificial Neural Network models. In formulating the Artificial Neural Network based predictive model, three layered networks have been constructed with sigmoid non-linearity. The models under study are different in the number of hidden neurons. After a thorough training and test procedure, neural net with three nodes in the hidden layer is found to be the ...

Chattopadhyay, Surajit

2006-01-01

237

In the present research, possibility of predicting average summer-monsoon rainfall over India has been analyzed through Artificial Neural Network models. In formulating the Artificial Neural Network based predictive model, three layered networks have been constructed with sigmoid non-linearity. The models under study are different in the number of hidden neurons. After a thorough training and test procedure, neural net with three nodes in the hidden layer is found to be the best predictive model.

Chattopadhyay, S

2006-01-01

238

Generalization ability of perceptrons with continuous outputs

In this paper we examine the influence of different input-output relations on the generalization ability of a single-layer perceptron. The input-output relations can be linear, binary, or sigmoid. With this choice we take into account most of the cases which are of present interest. The generalization problem will be realizable or unrealizable if the input-output relations for teacher and student are identical or not. We show that sometimes it can have a positive effect on the generalization ability, if one learns with errors.

Bös, S.; Kinzel, W.; Opper, M.

1993-02-01

239

Neural Network Aided Evaluation of Landslide Susceptibility in Southern Italy

Landslide hazard mapping is often performed through the identification and analysis of hillslope instability factors. In heuristic approaches, these factors are rated by the attribution of scores based on the assumed role played by each of them in controlling the development of a sliding process. The objective of this research is to forecast landslide susceptibility through the application of Artificial Neural Networks. In particular, given the availability of past events data, we mainly focused on the Calabria region (Italy). Vectors of eight hillslope factors (features) were considered for each considered event in this area (lithology, permeability, slope angle, vegetation cover in terms of type and density, land use, yearly rainfall and yearly temperature range). We collected 106 vectors and each one was labeled with its landslide susceptibility, which is assumed to be the output variable. Subsequently a set of these labeled vectors (examples) was used to train an artificial neural network belonging to the category of Multi-Layer Perceptron (MLP) to evaluate landslide susceptibility. Then the neural network predictions were verified on the vectors not used in the training (validation set), i.e. in previously unseen locations. The comparison between the expected output and the artificial neural network output showed satisfactory results, reporting a prediction discrepancy of less than 4.3%. This is an encouraging preliminary approach towards a systematic introduction of artificial neural network in landslide hazard assessment and mapping in the considered area.

Rampone, Salvatore; Valente, Alessio

240

Handwritten Farsi Character Recognition using Artificial Neural Network

Neural Networks are being used for character recognition from last many years but most of the work was confined to English character recognition. Till date, a very little work has been reported for Handwritten Farsi Character recognition. In this paper, we have made an attempt to recognize handwritten Farsi characters by using a multilayer perceptron with one hidden layer. The error backpropagation algorithm has been used to train the MLP network. In addition, an analysis has been carried out to determine the number of hidden nodes to achieve high performance of backpropagation network in the recognition of handwritten Farsi characters. The system has been trained using several different forms of handwriting provided by both male and female participants of different age groups. Finally, this rigorous training results an automatic HCR system using MLP network. In this work, the experiments were carried out on two hundred fifty samples of five writers. The results showed that the MLP networks trained by the err...

Ahangar, Reza Gharoie

2009-01-01

241

Recognition of Japanese finger spelling gestures using neural networks.

Effective communication with the hearing and speech impaired often requires at least a basic working knowledge of sign language gestures, without which a memo pad and pen, or a mobile phone's notepad is indispensable. The aim of this study was to build a neural network that could be used to recognize static finger-hand gestures of the yubimoji, the Japanese sign language syllabary. To build the network, signal inputs from a data glove interface were taken for each of the static yubimoji gestures. The network was trained and tested 10 times using a multilayer perceptron model. Overall, only 18 of the 41 static gestures were successfully recognized. One of the reasons was attributed to the inability of the data glove to measure gesture directions particularly for yubimoji gestures with similar finger configurations. Future work will focus on these problems as well as the inclusion of dynamic yubimoji gestures. PMID:20143958

Machacon, H T C; Shiga, S

2010-05-01

242

Feedforward neural network with adaptive reference pattern layer.

A hybrid neural network architecture is investigated for modeling purposes. The proposed hybrid is based on the multilayer perceptron (MLP) network. In addition to the usual hidden layers, the first hidden layer is selected to be an adaptive reference pattern layer. Each unit in this new layer incorporates a reference pattern that is located somewhere in the space spanned by the input variables. The outputs of these units are the component wise-squared differences between the elements of a reference pattern and the inputs. The reference pattern layer has some resemblance to the hidden layer of the radial basis function (RBF) networks. Therefore the proposed design can be regarded as a sort of hybrid of MLP and RBF networks. The presented benchmark experiments show that the proposed hybrid can provide significant advantages over standard MLPs and RBFs in terms of fast and efficient learning, and compact network structure. PMID:10401926

Lehtokangas, M

1999-02-01

243

Artificial neural networks in Space Station optimal attitude control

Innovative techniques of using "artificial neural networks" (ANN) for improving the performance of the pitch axis attitude control system of Space Station Freedom using control moment gyros (CMGs) are investigated. The first technique uses a feed-forward ANN with multi-layer perceptrons to obtain an on-line controller which improves the performance of the control system via a model following approach. The second technique uses a single layer feed-forward ANN with a modified back propagation scheme to estimate the internal plant variations and the external disturbances separately. These estimates are then used to solve two differential Riccati equations to obtain time varying gains which improve the control system performance in successive orbits.

Kumar, Renjith R.; Seywald, Hans; Deshpande, Samir M.; Rahman, Zia

1995-01-01

244

Neural network-based systems for handprint OCR applications.

Over the last five years or so, neural network (NN)-based approaches have been steadily gaining performance and popularity for a wide range of optical character recognition (OCR) problems, from isolated digit recognition to handprint recognition. We present an NN classification scheme based on an enhanced multilayer perceptron (MLP) and describe an end-to-end system for form-based handprint OCR applications designed by the National Institute of Standards and Technology (NIST) Visual Image Processing Group. The enhancements to the MLP are based on (i) neuron activations functions that reduce the occurrences of singular Jacobians; (ii) successive regularization to constrain the volume of the weight space; and (iii) Boltzmann pruning to constrain the dimension of the weight space. Performance characterization studies of NN systems evaluated at the first OCR systems conference and the NIST form-based handprint recognition system are also summarized. PMID:18276327

Ganis, M D; Wilson, C L; Blue, J L

1998-01-01

245

Directory of Open Access Journals (Sweden)

Full Text Available An electrocardiogram (ECG beat classification scheme based on multiple signal classification (MUSIC algorithm, morphological descriptors, and neural networks is proposed for discriminating nine ECG beat types. These are normal, fusion of ventricular and normal, fusion of paced and normal, left bundle branch block, right bundle branch block, premature ventricular concentration, atrial premature contraction, paced beat, and ventricular flutter. ECG signal samples from MIT-BIH arrhythmia database are used to evaluate the scheme. MUSIC algorithm is used to calculate pseudospectrum of ECG signals. The low-frequency samples are picked to have the most valuable heartbeat information. These samples along with two morphological descriptors, which deliver the characteristics and features of all parts of the heart, form an input feature vector. This vector is used for the initial training of a classifier neural network. The neural network is designed to have nine sample outputs which constitute the nine beat types. Two neural network schemes, namely multilayered perceptron (MLP neural network and a probabilistic neural network (PNN, are employed. The experimental results achieved a promising accuracy of 99.03% for classifying the beat types using MLP neural network. In addition, our scheme recognizes NORMAL class with 100% accuracy and never misclassifies any other classes as NORMAL.

2009-03-01

246

Neural Network Control of Asymmetrical Multilevel Converters

Directory of Open Access Journals (Sweden)

Full Text Available This paper proposes a neural implementation of a harmonic eliminationstrategy (HES to control a Uniform Step Asymmetrical Multilevel Inverter(USAMI. The mapping between the modulation rate and the requiredswitching angles is learned and approximated with a Multi-Layer Perceptron(MLP neural network. After learning, appropriate switching angles can bedetermined with the neural network leading to a low-computational-costneural controller which is well suited for real-time applications. Thistechnique can be applied to multilevel inverters with any number of levels. Asan example, a nine-level inverter and an eleven-level inverter are consideredand the optimum switching angles are calculated on-line. Comparisons to thewell-known sinusoidal pulse-width modulation (SPWM have been carriedout in order to evaluate the performance of the proposed approach. Simulationresults demonstrate the technical advantages of the proposed neuralimplementation over the conventional method (SPWM in eliminatingharmonics while controlling a nine-level and eleven-level USAMI. Thisneural approach is applied for the supply of an asynchronous machine andresults show that it ensures a highest quality torque by efficiently cancelingthe harmonics generated by the inverters.

Patrice WIRA

2009-12-01

247

Shale Gas reservoirs characterization using neural network

In this paper, a tentative of shale gas reservoirs characterization enhancement from well-logs data using neural network is established. The goal is to predict the Total Organic carbon (TOC) in boreholes where the TOC core rock or TOC well-log measurement does not exist. The Multilayer perceptron (MLP) neural network with three layers is established. The MLP input layer is constituted with five neurons corresponding to the Bulk density, Neutron porosity, sonic P wave slowness and photoelectric absorption coefficient. The hidden layer is forms with nine neurons and the output layer is formed with one neuron corresponding to the TOC log. Application to two boreholes located in Barnett shale formation where a well A is used as a pilot and a well B is used for propagation shows clearly the efficiency of the neural network method to improve the shale gas reservoirs characterization. The established formalism plays a high important role in the shale gas plays economy and long term gas energy production.

Ouadfeul, Sid-Ali; Aliouane, Leila

2014-05-01

248

Learning Dynamics of Photorefractive Neural Networks

This thesis investigates the optical implementation of neural networks utilizing dynamic photorefractive volume holography. The number of accessible degrees of freedom in a general holographic interconnection system is derived, and a cascaded-grating scheme that provides full, nondegenerate interconnections between two unsampled planes is presented. The dynamics of the formation of photorefractive volume holograms is considered. The impact of time-constant asymmetry on multiple hologram recording is evaluated. A basic framework for controlling the dynamics of photorefractive holograms is described and a number of dynamic copying methods for rejuvenating decayed holograms are identified. Experiments of linear dynamic copying using phase conjugation and nonlinear copying using an optical feedback loop are presented. The electrical fixing of photorefractive holograms in Sr _{0.75}Ba_{0.25 }Nb_2O_6 crystals is experimentally demonstrated and the physical mechanism is discussed. A number of neural learning algorithms are investigated for optical implementation. An Anti-Hebbian local learning algorithm is proposed to simplify the optical architecture of feedforward multilayer networks. Experimental demonstrations of several optical neural networks are presented. An optical perceptron is trained for face classification, and the use of dynamic copying for improving its performance is demonstrated. A two-layer network based on Kanerva's sparse, distributed memory model is implemented and trained for real-time handwritten character recognition. Finally an optical two-layer network for real-time face recognition, with moderate tolerance to shift, rotation, scale, and facial expression, is presented.

Qiao, Yong

249

Experimental characterization of the perceptron laser rangefinder

In this report, we characterize experimentally a scanning laser rangefinder that employs active sensing to acquire three-dimensional images. We present experimental techniques applicable to a wide variety of laser scanners, and document the results of applying them to a device manufactured by Perceptron. Nominally, the sensor acquires data over a 60 deg x 60 deg field of view in 256 x 256 pixel images at 2 Hz. It digitizes both range and reflectance pixels to 12 bits, providing a maximum range of 40 m and a depth resolution of 1 cm. We present methods and results from experiments to measure geometric parameters including the field of view, angular scanning increments, and minimum sensing distance. We characterize qualitatively problems caused by implementation flaws, including internal reflections and range drift over time, and problems caused by inherent limitations of the rangefinding technology, including sensitivity to ambient light and surface material. We characterize statistically the precision and accuracy of the range measurements. We conclude that the performance of the Perceptron scanner does not compare favorably with the nominal performance, that scanner modifications are required, and that further experimentation must be conducted.

Kweon, I. S.; Hoffman, Regis; Krotkov, Eric

1991-01-01

250

Energy Technology Data Exchange (ETDEWEB)

This paper presents a methodology for electrical load forecasting, using radial base functions as activation function in artificial neural networks with the training by backpropagation algorithm. This methodology is applied to short term electrical load forecasting (24 h ahead). Therefore, results are presented analyzing the use of radial base functions substituting the sigmoid function as activation function in multilayer perceptron neural networks. However, the main contribution of this paper is the proposal of a new formulation of load forecasting dedicated to the forecasting in several points of the electrical network, as well as considering several types of users (residential, commercial, industrial). It deals with the MLF (Multimodal Load Forecasting), with the same processing time as the GLF (Global Load Forecasting). (author)

Altran, A.B.; Lotufo, A.D.P.; Minussi, C.R. [Universidade Estadual Paulista Julio de Mesquita Filho (UNESP), Ilha Solteira, SP (Brazil). Dept. de Engenharia Eletrica], Emails: lealtran@yahoo.com.br, annadiva@dee.feis.unesp.br, minussi@dee.feis.unesp.br; Lopes, M.L.M. [Universidade Estadual Paulista Julio de Mesquita Filho (UNESP), Ilha Solteira, SP (Brazil). Dept. de Matematica], E-mail: mara@mat.feis.unesp.br

2009-07-01

251

Scientific Electronic Library Online (English)

Full Text Available SciELO Mexico | Language: Spanish Abstract in spanish El presente artículo describe la implementación de un sistema de predicción de fallos en redes LAN (fallos de timeout y rechazo en las conexiones), utilizando redes neuronales artificiales Perceptrón Multicapa. Se describe como se implementó el sistema, las pruebas realizadas para la selección de lo [...] s parámetros propios de la red neuronal, como del algoritmo de entrenamiento y los resultados de evaluación obtenidos. Abstract in english The paper presents the implementation of a system for predicting failures in LAN (timeout failure and rejection of connections), using neural networks (multilayer perceptron). It describes the implementation of the system, experiments conducted for the selection of specific parameters of the neural [...] network, training algorithm and evaluation results.

Gustavo A., García; Octavio, Salcedo.

2010-06-01

252

Scientific Electronic Library Online (English)

Full Text Available SciELO Mexico | Language: Spanish Abstract in spanish El presente artículo describe la implementación de un sistema de predicción de fallos en redes LAN (fallos de timeout y rechazo en las conexiones), utilizando redes neuronales artificiales Perceptrón Multicapa. Se describe como se implementó el sistema, las pruebas realizadas para la selección de lo [...] s parámetros propios de la red neuronal, como del algoritmo de entrenamiento y los resultados de evaluación obtenidos. Abstract in english The paper presents the implementation of a system for predicting failures in LAN (timeout failure and rejection of connections), using neural networks (multilayer perceptron). It describes the implementation of the system, experiments conducted for the selection of specific parameters of the neural [...] network, training algorithm and evaluation results.

Gustavo A., García; Octavio, Salcedo.

253

Directory of Open Access Journals (Sweden)

Full Text Available In this study, the use of artificial neural network (ANN based model, multi-layer perceptron (MLP network, to compute the transfer capabilities in a multi-area power system was explored. The input for the ANN is load status and the outputs are the transfer capability among the system areas, voltage magnitudes and voltage angles at concerned buses of the areas under consideration. The repeated power flow (RPF method is used in this paper for calculating the power transfer capability, voltage magnitudes and voltage angles necessary for the generation of input-output patterns for training the proposed MLP neural network. Preliminary investigations on a three area 30-bus system reveal that the proposed model is computationally faster than the conventional method.

Palukuru NAGENDRA

2010-12-01

254

Learning from correlated patterns by simple perceptrons

International Nuclear Information System (INIS)

Learning behavior of simple perceptrons is analyzed for a teacher-student scenario in which output labels are provided by a teacher network for a set of possibly correlated input patterns, and such that the teacher and student networks are of the same type. Our main concern is the effect of statistical correlations among the input patterns on learning performance. For this purpose, we extend to the teacher-student scenario a methodology for analyzing randomly labeled patterns recently developed in Shinzato and Kabashima 2008 J. Phys. A: Math. Theor. 41 324013. This methodology is used for analyzing situations in which orthogonality of the input patterns is enhanced in order to optimize the learning performance

255

[Application of artificial neural networks on the prediction of surface ozone concentrations].

Ozone is an important secondary air pollutant in the lower atmosphere. In order to predict the hourly maximum ozone one day in advance based on the meteorological variables for the Wanqingsha site in Guangzhou, Guangdong province, a neural network model (Multi-Layer Perceptron) and a multiple linear regression model were used and compared. Model inputs are meteorological parameters (wind speed, wind direction, air temperature, relative humidity, barometric pressure and solar radiation) of the next day and hourly maximum ozone concentration of the previous day. The OBS (optimal brain surgeon) was adopted to prune the neutral work, to reduce its complexity and to improve its generalization ability. We find that the pruned neural network has the capacity to predict the peak ozone, with an agreement index of 92.3%, the root mean square error of 0.0428 mg/m3, the R-square of 0.737 and the success index of threshold exceedance 77.0% (the threshold O3 mixing ratio of 0.20 mg/m3). When the neural classifier was added to the neural network model, the success index of threshold exceedance increased to 83.6%. Through comparison of the performance indices between the multiple linear regression model and the neural network model, we conclud that that neural network is a better choice to predict peak ozone from meteorological forecast, which may be applied to practical prediction of ozone concentration. PMID:22619942

Shen, Lu-Lu; Wang, Yu-Xuan; Duan, Lei

2011-08-01

256

Mathematically Reduced Chemical Reaction Mechanism Using Neural Networks

Energy Technology Data Exchange (ETDEWEB)

This is the final technical report for the project titled 'Mathematically Reduced Chemical Reaction Mechanism Using Neural Networks'. The aim of the project was to develop an efficient chemistry model for combustion simulations. The reduced chemistry model was developed mathematically without the need of having extensive knowledge of the chemistry involved. To aid in the development of the model, Neural Networks (NN) was used via a new network topology known as Non-linear Principal Components Analysis (NPCA). A commonly used Multilayer Perceptron Neural Network (MLP-NN) was modified to implement NPCA-NN. The training rate of NPCA-NN was improved with the GEneralized Regression Neural Network (GRNN) based on kernel smoothing techniques. Kernel smoothing provides a simple way of finding structure in data set without the imposition of a parametric model. The trajectory data of the reaction mechanism was generated based on the optimization techniques of genetic algorithm (GA). The NPCA-NN algorithm was then used for the reduction of Dimethyl Ether (DME) mechanism. DME is a recently discovered fuel made from natural gas, (and other feedstock such as coal, biomass, and urban wastes) which can be used in compression ignition engines as a substitute for diesel. An in-house two-dimensional Computational Fluid Dynamics (CFD) code was developed based on Meshfree technique and time marching solution algorithm. The project also provided valuable research experience to two graduate students.

Ziaul Huque

2007-08-31

257

Modified TAG neural network for large-scale optical implementation

Training by adaptive gain (TAG) neural network model, which had been developed for optical implementation of large-scale artificial neural networks, is further extended for better performance and its feasibility is demonstrated by a small-scale electro-optic implementation. For fully interconnected single-layer neural networks with N input and M output neurons the modified TAG model contains two different types of interconnections, i.e., MN fixed global interconnections and (beta) N + M adaptive local interconnections. For the original TAG model the number of adaptive local interconnections (beta) was set to 1, and the interconnections were understood as adaptive gain. For 2-dimensional input and output patterns the fixed global interconnections may be achieved by page-oriented holograms, and the adaptive local interconnections by spatial light modulators. The original and modified TAG models require much less adaptive elements than the popular perceptron model with fully adaptive global interconnections, and provide possibilities of implementing large-scale artificial neural networks with some sacrifice in performance. The training algorithm is based on gradient descent and error back-propagation, and is easily extensible to multi-layer architecture. Computer simulation and electro-optic implementation demonstrate much better performance of the modified TAG model compared to the original TAG model.

Lee, Soo-Young; Lee, Hyeuk-Jae; Shin, Sang-Yung

1992-10-01

258

This paper presents a comparison of the performances of neural network and linear predictors for near-lossless compression of EEG signals. Three neural network predictors, namely, single-layer perceptron (SLP), multilayer perceptron (MLP), and Elman network (EN), and two linear predictors, namely, autoregressive model (AR) and finite-impulse response filter (FIR) are used. For all the predictors, uniform quantization is applied on the residue signals obtained as the difference between the original and the predicted values. The maximum allowable reconstruction error delta is varied to determine the theoretical bound delta 0 for near-lossless compression and the corresponding bit rate rp. It is shown that among all the predictors, the SLP yields the best results in achieving the lowest values for delta 0 and rp. The corresponding values of the fidelity parameters, namely, percent of root-mean-square difference, peak SNR and cross correlation are also determined. A compression efficiency of 82.8% is achieved using the SLP with a near-lossless bound delta 0 = 3, with the diagnostic quality of the reconstructed EEG signal preserved. Thus, the proposed near-lossless scheme facilitates transmission of real time as well as offline EEG signals over network to remote interpretation center economically with less bandwidth utilization compared to other known lossless and near-lossless schemes. PMID:18270040

Sriraam, N; Eswaran, C

2008-01-01

259

Dynamic Baysesian state-space model with a neural network for an online river flow prediction

The usefulness of artificial neural networks in complex hydrological modeling has been demonstrated by successful applications. Several different types of neural network have been used for the hydrological modeling task but the multi-layer perceptron (MLP) neural network (also known as the feed-forward neural network) has enjoyed a predominant position because of its simplicity and its ability to provide good approximations. In many hydrological applications of MLP neural networks, the gradient descent-based batch learning algorithm such as back-propagation, quasi-Newton, Levenburg-Marquardt, and conjugate gradient algorithms has been used to optimize the cost function (usually by minimizing the error function in the prediction) by updating the parameters and structure in a neural network defined using a set of input-output training examples. Hydrological systems are highly with time-varying inputs and outputs, and are characterized by data that arrive sequentially. The gradient descent-based batch learning approaches that are implemented in MLP neural networks have significant disadvantages for online dynamic hydrological modeling because they could not update the model structure and parameter when a new set of hydrological measurement data becomes available. In addition, a large amount of training data is always required off-line with a long model training time. In this work, a dynamic nonlinear Bayesian state-space model with a multi-layer perceptron (MLP) neural network via a sequential Monte Carlo (SMC) learning algorithm is proposed for an online dynamic hydrological modeling. This proposed new method of modeling is herein known as MLP-SMC. The sequential Monte Carlo learning algorithm in the MLP-SMC is designed to evolve and adapt the weight of a MLP neural network sequentially in time on the arrival of each new item of hydrological data. The weight of a MLP neural network is treated as the unknown dynamic state variable in the dynamic Bayesian state-space model formulation. The nonlinear Monte Carlo filtering algorithm is based on recursively constructing the posterior probability density (distribution) of the state variable of neural network's weight, with respect to measured data (in our case, river flow), through a random trajectory of the state by entities called 'particles' in the dynamic state-space model formulation. A weight, which is the probability of the trajectory of the state, is assigned to each particle by a Bayesian correction term based on measurement. The algorithms differ in the way that the swarm of particles evolves and adapts to incoming online measurement data. In order to demonstrate the efficiency and usefulness of the proposed MLP-SMC, a practical application of hydrological modeling is carried out to predict the river flow sequentially in advance on the arrival of each new item of river flow data at intervals of 10 minutes. The performance of the proposed MLP-SMC is compared with the performance of a multi-layer perceptron (MLP) model trained using the back-propagation learning algorithm (MLP-BP) in which a batch off-line learning algorithm is implemented. The results show that the proposed MLP-SMC shows superiority in terms of model accuracy and computational cost compared with MLP-BP. The sequential Monte Carlo learning algorithm implemented in MLP-SMC is shown to have less sensitivity to noisy and sparsely distributed data compared to the batch off-line learning algorithm used in MLP-BP.

Ham, Jonghwa; Hong, Yoon-Seok

2013-04-01

260

Adaptive neural network vector predictor

In this paper, an adaptive neural network vector predictor is designed in order to improve the performance of the predictive component of the predictive vector quantizer (PVQ). The proposed vector predictor consists of a set of dedicated predictors (experts) where each predictor is optimized for a particular class of input vectors. In our simulations, we used five multi-layer perceptrons (MLP) to design our expert predictors. Each MLP predictor is separately trained by using a set of training vectors that belong to a particular class. The class identity of each training vector is determined by its directional variances. In our current implementation, one predictor is optimized for stationary blocks and four other predictors are designed for horizontal, vertical, 45 degree and 135 degree diagonally oriented edge blocks. The back-propagation algorithm is used for training each network. The directional variances of the neighboring blocks are used to select the appropriate expert predictor for the current input block. Therefore, no overhead information is transmitted in order to inform the receiver about the predictor selection. Our simulation shows that the proposed scheme gives an improvement of more than 1 dB over the predictor consisting of a single MLP predictor. The perceptual quality of the predicted images are also significantly improved.

Wang, Lin-Cheng; Rizvi, Syed A.; Nasrabadi, Nasser M.; Mirelli, Vincent

1996-03-01

261

Training Dynamics and Neural Network Performance.

We use an analysis of a simple model of recurrent network dynamics to gain qualitative insights into the training dynamics of feedforward multilayer perceptrons (MLPs) used for classification. These insights suggest changes to the training methods used for MLPs that improve network performance significantly. In previous work, the probabilistic neural network (PNN) was shown to provide better zero-reject error performance on character and fingerprint classification problems than radial basis function and MLP-based neural network methods. We will show that performance equal to or better than PNN can be achieved with a single three-layer MLP by making fundamental changes in the network optimization strategy. These changes are: 1) use of neuron activation functions, which reduce the probability of singular Jacobians; 2) use of successive regularization to constrain the volume of the minimized weight space; 3) use of Boltzmann pruning to constrain the dimension of the weight space; 4) use of Prior class probabilities to normalize all error calculations, so that statistically significant samples of rare but important classes can be included without distorting the error surface. All four of these changes are made in the inner loop of a conjugate gradient optimization iteration and are intended to simplify the training dynamics of the optimization. On handprinted digits and fingerprint classification problems these modifications improve error-reject performance by factors between 2 and 4, and reduce network size by 40-60%. Copyright 1997 Elsevier Science Ltd. PMID:12662879

Omidvar, Omid M.; Blue, James L.; Wilson, Charles L.

1997-07-01

262

Prospecting droughts with stochastic artificial neural networks

SummaryA non-linear multivariate model based on an artificial neural network multilayer perceptron is presented, that includes a random component. The developed model is applied to generate monthly streamflows, which are used to obtain synthetic annual droughts. The calibration of the model was undertaken using monthly streamflow records of several geographical sites of a basin. The model calibration consisted of training the neural network with the error back-propagation learning algorithm, and adding a normally distributed random noise. The model was validated by comparing relevant statistics of synthetic streamflow series to those of historical records. Annual droughts were calculated from the generated streamflow series, and then the expected values of length, intensity and magnitude of the droughts were assessed. An exercise on identical basis was made applying a second order auto-regressive multivariate model, AR(2), to compare its results with those of the developed model. The proposed model outperforms the AR(2) model in reproducing the future drought scenarios.

Ochoa-Rivera, Juan Camilo

2008-04-01

263

A novel algorithm for tunable compression to within the precision of reproduction targets, or storage, is proposed. The new algorithm is termed the `Perceptron Algorithm', which utilises simple existing concepts in a novel way, has multiple immediate commercial application aspects as well as it opens up a multitude of fronts in computational science and technology. The aims of this paper are to present the concepts underlying the algorithm, observations by its application to some example cases, and the identification of a multitude of potential areas of applications such as: image compression by orders of magnitude, signal compression including sound as well, image analysis in a multilayered detailed analysis, pattern recognition and matching and rapid database searching (e.g. face recognition), motion analysis, biomedical applications e.g. in MRI and CAT scan image analysis and compression, as well as hints on the link of these ideas to the way how biological memory might work leading to new points of view i...

Vassiliadis, V S

2006-01-01

264

Unfolding the neutron spectrum of a NE213 scintillator using artificial neural networks

International Nuclear Information System (INIS)

Artificial neural networks technology has been applied to unfold the neutron spectra from the pulse height distribution measured with NE213 liquid scintillator. Here, both the single and multi-layer perceptron neural network models have been implemented to unfold the neutron spectrum from an Am-Be neutron source. The activation function and the connectivity of the neurons have been investigated and the results have been analyzed in terms of the network's performance. The simulation results show that the neural network that utilizes the Satlins transfer function has the best performance. In addition, omitting the bias connection of the neurons improve the performance of the network. Also, the SCINFUL code is used for generating the response functions in the training phase of the process. Finally, the results of the neural network simulation have been compared with those of the FORIST unfolding code for both 241Am-Be and 252Cf neutron sources. The results of neural network are in good agreement with FORIST code.

265

An Oil Fraction Neural Sensor Developed Using Electrical Capacitance Tomography Sensor Data

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents novel research on the development of a generic intelligent oil fraction sensor based on Electrical Capacitance Tomography (ECT data. An artificial Neural Network (ANN has been employed as the intelligent system to sense and estimate oil fractions from the cross-sections of two-component flows comprising oil and gas in a pipeline. Previous works only focused on estimating the oil fraction in the pipeline based on fixed ECT sensor parameters. With fixed ECT design sensors, an oil fraction neural sensor can be trained to deal with ECT data based on the particular sensor parameters, hence the neural sensor is not generic. This work focuses on development of a generic neural oil fraction sensor based on training a Multi-Layer Perceptron (MLP ANN with various ECT sensor parameters. On average, the proposed oil fraction neural sensor has shown to be able to give a mean absolute error of 3.05% for various ECT sensor sizes.

Khursiah Zainal-Mokhtar

2013-08-01

266

Unfolding the neutron spectrum of a NE213 scintillator using artificial neural networks.

Artificial neural networks technology has been applied to unfold the neutron spectra from the pulse height distribution measured with NE213 liquid scintillator. Here, both the single and multi-layer perceptron neural network models have been implemented to unfold the neutron spectrum from an Am-Be neutron source. The activation function and the connectivity of the neurons have been investigated and the results have been analyzed in terms of the network's performance. The simulation results show that the neural network that utilizes the Satlins transfer function has the best performance. In addition, omitting the bias connection of the neurons improve the performance of the network. Also, the SCINFUL code is used for generating the response functions in the training phase of the process. Finally, the results of the neural network simulation have been compared with those of the FORIST unfolding code for both (241)Am-Be and (252)Cf neutron sources. The results of neural network are in good agreement with FORIST code. PMID:19586776

Sharghi Ido, A; Bonyadi, M R; Etaati, G R; Shahriari, M

2009-10-01

267

Unfolding the neutron spectrum of a NE213 scintillator using artificial neural networks

Energy Technology Data Exchange (ETDEWEB)

Artificial neural networks technology has been applied to unfold the neutron spectra from the pulse height distribution measured with NE213 liquid scintillator. Here, both the single and multi-layer perceptron neural network models have been implemented to unfold the neutron spectrum from an Am-Be neutron source. The activation function and the connectivity of the neurons have been investigated and the results have been analyzed in terms of the network's performance. The simulation results show that the neural network that utilizes the Satlins transfer function has the best performance. In addition, omitting the bias connection of the neurons improve the performance of the network. Also, the SCINFUL code is used for generating the response functions in the training phase of the process. Finally, the results of the neural network simulation have been compared with those of the FORIST unfolding code for both {sup 241}Am-Be and {sup 252}Cf neutron sources. The results of neural network are in good agreement with FORIST code.

Sharghi Ido, A. [Radiation Application Department, Shahid Beheshti University, Tehran (Iran, Islamic Republic of); Bonyadi, M.R. [Electrical and Computer Engineering Faculty, Shahid Beheshti University, Tehran (Iran, Islamic Republic of); Etaati, G.R. [Nuclear Engineering and Physics Faculty, Amir Kabir University of Technology, Tehran (Iran, Islamic Republic of); Shahriari, M. [Radiation Application Department, Shahid Beheshti University, Tehran (Iran, Islamic Republic of)], E-mail: m-shahriari@sbu.ac.ir

2009-10-15

268

Contextual classification methods, which require the extraction of complex spatial information over a range of scales, from fine details in local areas to large features that extend across the image, are necessary in many remote sensing image classification studies. This work presents a supervised adaptive object recognition model which integrates scale-space filtering techniques for feature extraction within a neural classification procedure based on multilayer perceptron (MLP). The salient aspect of the model is the integration within the back-propagation learning task of the search of the most adequate filter parameters. The experimental evaluation of the method has been conducted coping with object recognition in high-resolution remote sensing imagery. To investigate whether the strategy can be considered an alternative to conventional procedures the results were compared with those obtained by a well known contextual classification scheme.

Binaghi, Elisabetta; Gallo, Ignazio; Pepe, Monica

2003-03-01

269

Background True date palms (Phoenix dactylifera L.) are impressive trees and have served as an indispensable source of food for mankind in tropical and subtropical countries for centuries. The aim of this study is to differentiate date palm tree varieties by analysing leaflet cross sections with technical/optical methods and artificial neural networks (ANN). Results Fluorescence microscopy images of leaflet cross sections have been taken from a set of five date palm tree cultivars (Hewlat al Jouf, Khlas, Nabot Soltan, Shishi, Um Raheem). After features extraction from images, the obtained data have been fed in a multilayer perceptron ANN with backpropagation learning algorithm. Conclusions Overall, an accurate result in prediction and differentiation of date palm tree cultivars was achieved with average prediction in tenfold cross-validation is 89.1% and reached 100% in one of the best ANN. PMID:24564551

2014-01-01

270

Directory of Open Access Journals (Sweden)

Full Text Available In this contribution, human face as biometric is considered. Original method of feature extraction from image data is introduced using MLP (multilayer perceptron and PCA (principal component analysis. This method is used in human face recognition system and results are compared to face recognition system using PCA directly, to a system with direct classification of input images by MLP and RBF (radial basis function networks, and to a system using MLP as a feature extractor and MLP and RBF networks in the role of classifier. Also a two-stage method for face recognition is presented, in which Kohonen self-organizing map is used as a feature extractor. MLP and RBF network are used as classifiers. In order to obtain deeper insight into presented methods, also visualizations of internal representation of input data obtained by neural networks are presented.

J. Pavlovicova

2007-04-01

271

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents a hybrid approach with two phases for improving the performance of training artificial neural networks (ANNs by selection of the most important instances for training, and then reduction the dimensionality of features. The ANNs which are applied in this paper for validation, are included Multi-Layer Perceptron (MLP and Neuro-Fuzzy Network (NFN. In the first phase, the Modified Fast Condensed Nearest Neighbor (MFCNN algorithm is used to construct the subset with instances very close to the decision boundary. It leads to achieve the instances more useful for training the network. And in the second phase, an Ant-based approach to the supervised reduction of feature dimensionality is introduced, aims to reduce the complexity, and improve the accuracy of learning the ANN. The main purpose of this method is to enhance the classification performance by improving the quality of the training set. Experimental results illustrated the efficiency of the proposed approach.

Ali Abroudi

2013-04-01

272

Static sign language recognition using 1D descriptors and neural networks

A frame work for static sign language recognition using descriptors which represents 2D images in 1D data and artificial neural networks is presented in this work. The 1D descriptors were computed by two methods, first one consists in a correlation rotational operator.1 and second is based on contour analysis of hand shape. One of the main problems in sign language recognition is segmentation; most of papers report a special color in gloves or background for hand shape analysis. In order to avoid the use of gloves or special clothing, a thermal imaging camera was used to capture images. Static signs were picked up from 1 to 9 digits of American Sign Language, a multilayer perceptron reached 100% recognition with cross-validation.

Solís, José F.; Toxqui, Carina; Padilla, Alfonso; Santiago, César

2012-10-01

273

Dynamic model of a PEM electrolyser based on artificial neural networks

Energy Technology Data Exchange (ETDEWEB)

Hydrogen production by electrolysis is emerging as a promising way to meet future fuel demand, and developing models capable of simulating the operation of electrolysis devices is indispensable to efficiently design power generation systems, reduce manufacturing costs and save resources. The nonlinear nature of the Artificial Neural Network (ANN) plays a key role in developing models predicting the performance of complex systems. The behaviour of a Polymer Electrolyte Membrane (PEM) Electrolyser of three cell stack was modelled successfully using a Multilayer Perceptron Network (MLP). This dynamic model was trained to learn the internal relationships of this electrolysis device and predict its behaviour without physical equations. Electric current supply and operation temperature were used as input vector able to predict each cell voltage behaviour. An accuracy (< 2%) was reached after comparing the single cell performance of the real electrolyser versus the ANN based model. This predictive model can be used as a virtual device into a more complex energy system.

Chavez-Ramirez, A.U.; Munoz-Guerrero, R.; Sanchez-Huerta, V.; Ramirez-Arredondo, Juan M.; Ornelas, R.; Arriaga, L.G.; Siracusano, S.; Brunaccini, G.; Napoli, G.; Antonucci, V.; Arico, A.S.

2011-01-15

274

Fast converging minimum probability of error neural network receivers for DS-CDMA communications

In this work we consider the problem of detecting the information bit of a direct-sequence code-division-multiple-access (DS-CDMA) user in the presence of spread spectrum interference and AWGN using a multi-layer perceptron neural network receiver. We develop a fast converging adaptive training algorithm that minimizes the mean square error (MSE) at the output of the receiver. The proposed adaptive algorithm has two key features: (i) it utilizes constraints that are derived from properties of the optimum single-user decision boundary for AWGN multiple-access channels, and (ii) it embeds importance sampling principles directly into the receiver optimization process. Simulation studies illustrate the BER performance of the proposed scheme.

Matyjas, John D.; Psaromiligkos, Ioannis N.; Batalama, Stella N.; Medley, Michael J.

2003-07-01

275

Early detection of incipient faults in power plants using accelerated neural network learning

International Nuclear Information System (INIS)

An important aspect of power plant automation is the development of computer systems able to detect and isolate incipient (slowly developing) faults at the earliest possible stages of their occurrence. In this paper, the development and testing of such a fault detection scheme is presented based on recognition of sensor signatures during various failure modes. An accelerated learning algorithm, namely adaptive backpropagation (ABP), has been developed that allows the training of a multilayer perceptron (MLP) network to a high degree of accuracy, with an order of magnitude improvement in convergence speed. An artificial neural network (ANN) has been successfully trained using the ABP algorithm, and it has been extensively tested with simulated data to detect and classify incipient faults of various types and severity and in the presence of varying sensor noise levels

276

Use of artificial neural networks for prognosis of charcoal prices in Minas Gerais

Directory of Open Access Journals (Sweden)

Full Text Available Energy is an important factor of economic growth and is critical to the stability of a nation. Charcoal is a renewable energy resource and is a fundamental input to the development of the Brazilian forest-based industry. The objective of this study is to provide a prognosis of the charcoal price series for the year 2007 by using Artificial Neural Networks. A feedforward multilayer perceptron ANN was used, the results of which are close to reality. The main findings are that: real prices of charcoal dropped between 1975 and 2000 and rose from the early 21st century; the ANN with two hidden layers was the architecture making the best prediction; the most effective learning rate was 0.99 and 600 cycles, representing the most satisfactory and accurate ANN training. Prediction using ANN was found to be more accurate when compared by the mean squared error to other studies modeling charcoal price series in Minas Gerais state.

Luiz Moreira Coelho Junior

2013-06-01

277

Directory of Open Access Journals (Sweden)

Full Text Available The main goal of the presented work is to analyse the performance of the Multi-Layer Perceptron (MLP neural network for flow regime classification based on sets of simulated Electrical Capacitance Tomography (ECT data. Normalised ECT data have been used to separately train several MLPs employing various commonly used back-propagation learning algorithms, namely the Levenberg-Marquardt (LM, Quasi-Newton (QN and Resilient-Backpropagation (RP, to classify the gas-oil flow regimes. The performances of the MLPs have been analysed based on their correct classification percentage (CCP. The results demonstrate the feasibility of using MLP, and the superiority of LM algorithm for flow regime classification based on ECT data.

Khursiah Zainal-Mokhtar

2009-02-01

278

Modeling mechanical properties of cast aluminum alloy using artificial neural network

International Nuclear Information System (INIS)

Modeling is widely used to investigate the mechanical properties of engineering materials due to increasing demand of low cost and high strength to weight ratio for many engineering applications. The aluminum casting alloys are cost competitive material and possess the desired properties. The mechanical properties largely depend upon composition of alloys and their processing method. Alloy design involves controlling mechanical properties via optimization of the composition and processing parameters. For optimization the possible root is empirical modeling and its more refined version is the analysis of the wide range of data using ANN (Artificial Neural Networks) modeling. The modeling of mechanical properties of the aluminum alloys are the main objective of present work. For this purpose, some data were collected and experimentally prepared using conventional casting method. A MLP (Multilayer Perceptron) network was developed, which is trained by using the error back propagation algorithm. (author)

279

Directory of Open Access Journals (Sweden)

Full Text Available In this study, an Artificial Neural Network (ANN model has been developed for aqueous solutions of electrolyte systems. Multilayer Perceptron (MLP and Radial Basis Function (RBF networks were applied to estimate vapor-liquid equilibrium data for ternary system of NH3-CO2-H2O. Experimental data, taken from the literature were divided into three sections of training, validating and testing. Mean Absolute Errors (MAE of the networks for training set are used as network selection criterion and to find optimal design of the networks. The performance of ANN models to predict partial and total pressures of NH3-CO2-H2O system were evaluated by comparing their results with the predictions of some thermodynamic models. The criterion for this comparison was the error between models perditions and the experimental data. The comparison indicated that both MLP and RBF models predict the system better than the thermodynamic models.

A. Ghaemi

2008-01-01

280

We describe the development of artificial neural networks (ANN) for the prediction of the properties of ceramic materials. The ceramics studied here include polycrystalline, inorganic, non-metallic materials and are investigated on the basis of their dielectric and ionic properties. Dielectric materials are of interest in telecommunication applications where they are used in tuning and filtering equipment. Ionic and mixed conductors are the subjects of a concerted effort in the search for new materials that can be incorporated into efficient, clean electrochemical devices of interest in energy production and greenhouse gas reduction applications. Multi-layer perceptron ANNs are trained using the back-propagation algorithm and utilise data obtained from the literature to learn composition-property relationships between the inputs and outputs of the system. The trained networks use compositional information to predict the relative permittivity and oxygen diffusion properties of ceramic materials. The results sh...

Scott, D J; Kilner, J A; Rossiny, J C H; McAlford, N N

2007-01-01

281

Artificial Neural Network Based Method to Mitigate Temporary Over-voltages

Directory of Open Access Journals (Sweden)

Full Text Available Uncontrolled energization of large power transformers may result in magnetizing inrush current of high amplitude and switching over-voltages. The most effective method for the limitation of the switching over-voltages is controlled switching since the magnitudes of the produced transients are strongly dependent on the closing instants of the switch.? We introduce a harmonic index that its minimum value is corresponding to the best-case switching time.? Also, this paper ?presents an Artificial Neural Network (ANN-based approach to ?estimate the optimum switching instants for real time applications. In the proposed ANN, second order Levenberg–Marquardt ? method is used to train the multilayer perceptron. ANN training is performed based on equivalent circuit parameters of the network. Thus, trained ANN is applicable to every studied system. To verify the effectiveness of the proposed index and accuracy of the ANN-based approach, two case studies are presented and demonstrated.

Iman Sadeghkhani

2011-09-01

282

Concept and design of the fast H1 second level trigger using artificial neural networks

Energy Technology Data Exchange (ETDEWEB)

The experiments at the HERA ep collider have to cope with machine background rates which exceed the rates of the majority ofinteresting physics interactions by several orders of magnitude. To deal with this the H1 experiment was designed with a four staged trigger system. The second trigger level is currently being equipped with a highly parallel computing system that is specialized to run Artificial Neural Network (ANN) algorithms. The trigger is based on CNAPS neurocomputer boards, which are able to compute multilayer perceptron networks with 64 inputs, 64 hidden nodes and one output node in less than 8 {mu}s. In this contribution we present the concept and design of this system. (author)

Kolanoski, H. [Humboldt-Universitaet, Berlin (Germany). Inst. fuer Physik; Getta, H.; Goldner, D. [Dortmund Univ. (Germany). Inst. fuer Physik] [and others

1996-07-01

283

Concept and design of the fast H1 second level trigger using artificial neural networks

International Nuclear Information System (INIS)

The experiments at the HERA ep collider have to cope with machine background rates which exceed the rates of the majority of interesting physics interactions by several orders of magnitude. To deal with this the H1 experiment was designed with a four staged trigger system. The second trigger level is currently being equipped with a highly parallel computing system that is specialized to run Artificial Neural Network (ANN) algorithms. The trigger is based on CNAPS neurocomputer boards, which are able to compute multilayer perceptron networks with 64 inputs, 64 hidden nodes and one output node in less than 8 ?s. In this contribution we present the concept and design of this system. (author)

284

Performance Evaluation of Neural Network Based Pulse-Echo Weld Defect Classifiers

Pulse-echo ultrasonic signal is used to detect weld defects with high probability. However, utilizing echo signal for defects classification is another issue that has attracted attention of many researchers who have devised algorithms and tested them against their own databases. In this paper, a study is conducted to score the performance of various algorithms against a single echo signal database. Algorithms tested the use of Wavelet Transform (WT), Fast Fourier Transform (FFT) and time domain echo signal features and employed several NN’s architectures such as Multi-Layer Perceptron Neural Network (MLP), Self Organizing Map (SOM) and others known to be good classifiers. The average performance of all can be viewed fair (90%) while some algorithms render success rate of about 94%. It seems that acquiring higher success rates out of a single fixed angle probe pulseecho set up needs new arrangements of data collection, which is under investigation.

Seyedtabaii, S.

2012-10-01

285

Functional Link Artificial Neural Network for Classification Task in Data Mining

Directory of Open Access Journals (Sweden)

Full Text Available In solving classification task of data mining, the traditional algorithm such as multi-layer perceptron takes longer time to optimize the weight vectors. At the same time, the complexity of the network increases as the number of layers increases. In this study, we have used Functional Link Artificial Neural Networks (FLANN for the task of classification. In contrast to multiple layer networks, FLANN architecture uses a single layer feed-forward network. Using the functionally expanded features FLANN overcomes the non-linearity nature of problems, which is commonly encountered in single layer networks. The features like simplicity of designing the architecture and low-computational complexity of the networks encourages us to use it in data mining task. An extensive simulation study is presented to demonstrate the effectiveness of the classifier.

B. B. Misra

2007-01-01

286

Neural Network Based Lna Design for Mobile Satellite Receiver

Directory of Open Access Journals (Sweden)

Full Text Available Paper presents a Neural Network Modelling approach to microwave LNA design. To acknowledge the specifications of the amplifier, Mobile Satellite Systems are analyzed. Scattering parameters of the LNA in the frequency range 0.5 to 18 GHz are calculated using a Multilayer Perceptron Artificial Neural Network model and corresponding smith charts and polar charts are plotted as output to the model. From these plots, the microwave scattering parameter description of the LNA are obtained. Model is efficiently trained using Agilent ATF 331M4 InGaAs/InP Low Noise pHEMT amplifier datasheet and the neural model’s output seem to follow the various device characteristic curves with high regression. Next, Maximum Allowable Gain and Noise figure of the device are modelled and plotted for the same frequency range. Finally, the optimized model is utilized as an interpolator and the resolution of the amplifying capability with noise characteristics are obtained for the L Band of MSS operation.

Abhijeet Upadhya

2014-08-01

287

Chaotic diagonal recurrent neural network

International Nuclear Information System (INIS)

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)

288

Evaluation of convolutional neural networks for visual recognition.

Convolutional neural networks provide an efficient method to constrain the complexity of feedforward neural networks by weight sharing and restriction to local connections. This network topology has been applied in particular to image classification when sophisticated preprocessing is to be avoided and raw images are to be classified directly. In this paper two variations of convolutional networks--neocognitron and a modification of neocognitron--are compared with classifiers based on fully connected feedforward layers (i.e., multilayer perceptron, nearest neighbor classifier, auto-encoding network) with respect to their visual recognition performance. Beside the original neocognitron a modification of the neocognitron is proposed which combines neurons from perceptron with the localized network structure of neocognitron. Instead of training convolutional networks by time-consuming error backpropagation, in this work a modular procedure is applied whereby layers are trained sequentially from the input to the output layer in order to recognize features of increasing complexity. For a quantitative experimental comparison with standard classifiers two very different recognition tasks have been chosen: handwritten digit recognition and face recognition. In the first example on handwritten digit recognition the generalization of convolutional networks is compared to fully connected networks. In several experiments the influence of variations of position, size, and orientation of digits is determined and the relation between training sample size and validation error is observed. In the second example recognition of human faces is investigated under constrained and variable conditions with respect to face orientation and illumination and the limitations of convolutional networks are discussed. PMID:18252491

Nebauer, C

1998-01-01

289

Prediction of the local power factor in BWR fuel cells by means of a multilayer neural network

International Nuclear Information System (INIS)

To the beginning of a new operation cycle in a BWR reactor the reactivity of this it increases by means of the introduction of fresh fuel, the one denominated reload fuel. The problem of the definition of the characteristics of this reload fuel represents a combinatory optimization problem that requires significantly a great quantity of CPU time for their determination. This situation has motivated to study the possibility to substitute the Helios code, the one which is used to generate the new cells of the reload fuel parameters, by an artificial neuronal network, with the purpose of predicting the parameters of the fuel reload cell of a BWR reactor. In this work the results of the one training of a multilayer neuronal net that can predict the local power factor (LPPF) in such fuel cells are presented. The prediction of the LPPF is carried out in those condition of beginning of the life of the cell (0.0 MWD/T, to 40% of holes in the one moderator, temperature of 793 K in the fuel and a moderator temperature of 560 K. The cells considered in the present study consist of an arrangement of 10x10 bars, of those which 92 contains U235, some of these bars also contain a concentration of Gd2O3 and 8 of them contain only water. The axial location inside the one assembles of recharge of these cells it is exactly up of the cells that contain natural uranium in the base of the reactor core. The training of the neuronal net is carried out by means of a retro-propagation algorithm that uses a space of training formed starting from previous evaluations of cells by means of the Helios code. They are also presented the results of the application of the neuronal net found for the prediction of the LPPF of some cells used in the real operation of the Unit One of the Laguna Verde Nuclear Power station. (Author)

290

PENGENALAN CITRA OBJEK SEDERHANA DENGAN JARINGAN SARAF TIRUAN METODE PERCEPTRON

Directory of Open Access Journals (Sweden)

Full Text Available Konsep bangunan dan benda-benda yang ada di sekeliling didasarkan dan dipengaruhi oleh konsep objek sederhana atau sering disebut geometri ruang tiga dimensi, yaitu memiliki panjang, lebar dan tinggi. Namun, dalam rancangan dan penggambarannya menggunakan gambar berdimensi dua saja. Sehingga pada konsep penggambarannya membutuhkan visualisasi yang lebih detail. Diharapkan jaringan syaraf tiruan metode perceptron dapat mengenali gambar yang sesuai dengan bentuk aslinya. Pada penelitian ini metode jaringan saraf yang digunakan adalah metode perceptron untuk mengenali citra objek sederhana. Objek yang digunakan yaitu bentuk bangun ruang yang terdiri dari kubus, kerucut, tabung, prisma, dan limas dengan berbagai jenisnya. Perangkat lunak yang digunakan pada pembuatan aplikasi ini adalah Borland Delphi 7.0. Dari hasil pelatihan dan pengujian jaringan saraf tiruan perceptron dapat mengenali pola dengan rata-rata 75,25 % dengan prosentase terendah yaitu 50,75 % dan prosentase tertinggi yaitu 92,65 %. Dengan prosentase yang cukup baik tersebut, sistem dapat digunakan untuk mengenali citra objek sederhana.

Ardi Pujiyanta

2012-05-01

291

Nonseparable data models for a single-layer perceptron

This paper describes two nonseparable data models that can be used to study the convergence properties of perceptron learning algorithms. A system identification formulation generates the training signal, with an input that is a zero-mean Gaussian random vector. One model is based on a two-layer perceptron configuration, while the second model has only one layer but with a multiplicative output node. The analysis in this paper focuses on Rosenblatt's training procedure, although the approach can be applied to other learning algorithms. Some examples of the performance surfaces are presented to illustrate possible convergence points of the algorithm for both nonseparable data models.

Shynk, John J.; Bershad, Neil J.

1992-07-01

292

Bionomic features of blowflies may be clarified and detailed by the deployment of appropriate modelling techniques such as artificial neural networks, which are mathematical tools widely applied to the resolution of complex biological problems. The principal aim of this work was to use three well-known neural networks, namely Multi-Layer Perceptron (MLP), Radial Basis Function (RBF), and Adaptive Neural Network-Based Fuzzy Inference System (ANFIS), to ascertain whether these tools would be able to outperform a classical statistical method (multiple linear regression) in the prediction of the number of resultant adults (survivors) of experimental populations of Chrysomya megacephala (F.) (Diptera: Calliphoridae), based on initial larval density (number of larvae), amount of available food, and duration of immature stages. The coefficient of determination (R2) derived from the RBF was the lowest in the testing subset in relation to the other neural networks, even though its R2 in the training subset exhibited virtually a maximum value. The ANFIS model permitted the achievement of the best testing performance. Hence this model was deemed to be more effective in relation to MLP and RBF for predicting the number of survivors. All three networks outperformed the multiple linear regression, indicating that neural models could be taken as feasible techniques for predicting bionomic variables concerning the nutritional dynamics of blowflies. PMID:20569135

Bianconi, Andre; Zuben, Claudio J. Von; Serapiao, Adriane B. de S.; Govone, Jose S.

2010-01-01

293

Directory of Open Access Journals (Sweden)

Full Text Available Epilepsy is a common neurological disorder that is characterized by recurrent unprovoked seizures. About 40 to 50 million people worldwide have epilepsy. In this paper the authors presents clinical decision support system (DSS for the diagnosis of epilepsy. The DSS is developed by using Multilayer Perceptron (MLP, Generalized Feed Forward Neural Network (GFF-NN and Elman Neural Network (E-NN. The validity of neural networks to diagnose the epilepsy is checked and the most suitable neural network is recommended for the diagnosis of epilepsy. Also the different feature enhancement techniques like principal component analysis (PCA, FFT and statistical parameters are used for the input dimensionality reduction. Epilepsy diagnosis is modeled as the classification of normal EEG, interictal EEG and ictal EEG. With the different input dimensionality reduction methods performance parameters of MLP, GFF-NN and E-NN are measured and compared. For the GFF-NN, number of free parameter is reduced up to 92.22% when PCA is used for input dimensionality reduction and its overall accuracy of is 98.61%.

KHARAT P.A and DUDUL S.V.

2011-12-01

294

Use of Neural Networks for Damage Assessment in a Steel Mast

DEFF Research Database (Denmark)

In this paper the possibility of using a Multilayer Perceptron (MLP) network trained with the Backpropagation Algorithm for detecting location and size of a damage in a civil engineering structure is investigated. The structure considered is a 20 m high steel lattice mast subjected to wind excitation. The basic idea is to train a neural network with simulated patterns of the relative changes in natural frequencies and corresponding sizes and locations of damages in order to recognize the behaviour of the damaged as well as the undamaged structure. Subjecting this trained neural network to measured values should imply information about damages states and locations. The training data are obtained by an FEM of the mast. Different damage scenarios are established by simulating a damage in one of the eight lower diagonals. The eight lower diagonals are cut and provided with bolted joints. Each bolted joint consists of 4 slice plates giving the possibilities of simulating a 1/4, 1/2, 3/4 and full reduction of the area of a diagonal. A damage is simulated by removing one or more splice plates in these bolted joints. The utility of the neural network approach is demonstrated by a simulation study as well as full-scale tests where the mast is identified by an ARMA-model. The results show that a neural network trained with simulated data is capable for detecting location of a damage in a steel lattice mast when the network is subjected to experimental data.·

Kirkegaard, Poul Henning; Rytter, A.

1995-01-01

295

Integrated on-line plant monitoring system for HTTR using neural networks

International Nuclear Information System (INIS)

The neural networks have been utilized in on-line monitoring-system of High Temperature Engineering Tested Reactor (HTTR) with thermal power of 30 MW. In this system, several neural networks can independently model the plant dynamics with different architecture, input and output signals and learning algorithm. Monitoring task of each neural network is also different, respectively. Those parallel method applications require distributed architecture of computer network for performing real-time tasks. One of main task is real-time monitoring by Multi-Layer Perceptron (MLP) in auto-associative mode, which can model and estimate the whole plant dynamics by training normal operational data only. The basic principle of the anomaly detection is to monitor the difference between process signals measured from the actual plant and the corresponding values estimated by MLP. Other tasks are on-line reactivity prediction, reactivity and helium leak monitoring, respectively. From the on-line test results, each neural network shows good prediction and reliable detection performances. (author)

296

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

Directory of Open Access Journals (Sweden)

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.

M. Agatonovi?

2012-12-01

297

Directory of Open Access Journals (Sweden)

Full Text Available This paper gives the definition of Transparent Neural Network “TNN” for the simulation of the global-local vision and its application to the segmentation of administrative document image. We have developed and have adapted a recognition method which models the contextual effects reported from studies in experimental psychology. Then, we evaluated and tested the TNN and the multi-layer perceptron “MLP”,which showed its effectiveness in the field of the recognition, in order to show that the TNN is clearer for the user and more powerful on the level of the recognition. Indeed, the TNN is the only system which makes it possible to recognize the document and its structure

Boulbaba Ben Ammar

2013-09-01

298

Equivalence between learning in noisy perceptrons and tree committee machines

We study learning from single presentation of examples (on-line learning) in single-layer perceptrons and tree committee machines (TCMs). Lower bounds for the perceptron generalization error as a function of the noise level ? in the teacher output are calculated. We find that local learning in a TCM with K hidden units is simply related to learning in a simple perceptron with a corresponding noise level ?(K). For a large number of examples and finite K the generalization error decays as ?-1CM, where ?CM is the number of examples per adjustable weight in the TCM. We also show that on-line learning is possible even in the K-->? limit, but with the generalization error decaying as ?-1/2CM. The simple Hebb rule can also be applied to the TCM, but now the error decays as ?-1/2CM for finite K and ?-1/4CM for K-->?. Exponential decay of the generalization error in both the noisy perceptron learning and in the TCM is obtained by using the learning by queries strategy.

Copelli, Mauro; Kinouchi, Osame; Caticha, Nestor

1996-06-01

299

Dilution in Boolean perceptrons that learn from noisy examples

We investigate the effect of dilution after learning on the generalization ability of single-layer Boolean perceptrons that learn from noisy examples. We present a thorough comparison between the relative performances of several well known learning rules. In particular, we show that the effect of dilution is always deleterious, and that the Bayes algorithm always gives the best generalization performance.

Barbato, D. M. L.; Fontanari, J. F.

1996-11-01

300

Parallel perceptrons (PPs) are very simple and efficient committee machines (a single layer of perceptrons with threshold activation functions and binary outputs, and a majority voting decision scheme), which nevertheless behave as universal approximators. The parallel delta (P-Delta) rule is an effective training algorithm, which, following the ideas of statistical learning theory used by the support vector machine (SVM), raises its generalization ability by maximizing the difference between the perceptron activations for the training patterns and the activation threshold (which corresponds to the separating hyperplane). In this paper, we propose an analytical closed-form expression to calculate the PPs' weights for classification tasks. Our method, called Direct Parallel Perceptrons (DPPs), directly calculates (without iterations) the weights using the training patterns and their desired outputs, without any search or numeric function optimization. The calculated weights globally minimize an error function which simultaneously takes into account the training error and the classification margin. Given its analytical and noniterative nature, DPPs are computationally much more efficient than other related approaches (P-Delta and SVM), and its computational complexity is linear in the input dimensionality. Therefore, DPPs are very appealing, in terms of time complexity and memory consumption, and are very easy to use for high-dimensional classification tasks. On real benchmark datasets with two and multiple classes, DPPs are competitive with SVM and other approaches but they also allow online learning and, as opposed to most of them, have no tunable parameters. PMID:21984498

Fernandez-Delgado, Manuel; Ribeiro, Jorge; Cernadas, Eva; Ameneiro, Senén Barro

2011-11-01

301

Analysis of JET charge exchange spectra using neural networks

International Nuclear Information System (INIS)

Active charge exchange spectra representing the local interaction of injected neutral beams and fully stripped impurity ions are hard to analyse due to strong blending with passive emission from the plasma edge. As a result, the deduced plasma parameters (e.g. ion temperature, rotation velocity, impurity density) cannot always be determined unambiguously. Also, the speed of the analysis is limited by the time consuming nonlinear least-squares minimization procedure. In practice, semi-manual analysis is necessary and fast, automatic analysis, based on currently used techniques, does not seem feasible. In this paper the development of a robust and accurate analysis procedure based on multi-layer perceptron (MLP) neural networks is described. This procedure is fully automatic and fast, thus enabling a real-time analysis of charge exchange spectra. Accuracy has been increased in several ways as compared to earlier straightforward neural network implementations and is comparable to a standard least-squares based analysis. Robustness is achieved by using a combination of different confidence measures. A novel technique for the creation of training data, suitable for high-dimensional inverse problems has been developed and used extensively. A new method for fast calculation of error bars directly from the hidden neurons in a MLP network is also described, and used as part of the confidence calculations. For demonstration purposes, a real-time ion temperature profile diagnostic based on this work has been implemented. (author)

302

Classification of Images Acquired with Colposcopy Using Artificial Neural Networks

OBJECTIVE To explore the advantages of using artificial neural networks (ANNs) to recognize patterns in colposcopy to classify images in colposcopy. PURPOSE Transversal, descriptive, and analytical study of a quantitative approach with an emphasis on diagnosis. The training test e validation set was composed of images collected from patients who underwent colposcopy. These images were provided by a gynecology clinic located in the city of Criciúma (Brazil). The image database (n = 170) was divided; 48 images were used for the training process, 58 images were used for the tests, and 64 images were used for the validation. A hybrid neural network based on Kohonen self-organizing maps and multilayer perceptron (MLP) networks was used. RESULTS After 126 cycles, the validation was performed. The best results reached an accuracy of 72.15%, a sensibility of 69.78%, and a specificity of 68%. CONCLUSION Although the preliminary results still exhibit an average efficiency, the present approach is an innovative and promising technique that should be deeply explored in the context of the present study. PMID:25374454

Simoes, Priscyla W; Izumi, Narjara B; Casagrande, Ramon S; Venson, Ramon; Veronezi, Carlos D; Moretti, Gustavo P; da Rocha, Edroaldo L; Cechinel, Cristian; Ceretta, Luciane B; Comunello, Eros; Martins, Paulo J; Casagrande, Rogerio A; Snoeyer, Maria L; Manenti, Sandra A

2014-01-01

303

Statistical process control using optimized neural networks: A case study.

The most common statistical process control (SPC) tools employed for monitoring process changes are control charts. A control chart demonstrates that the process has altered by generating an out-of-control signal. This study investigates the design of an accurate system for the control chart patterns (CCPs) recognition in two aspects. First, an efficient system is introduced that includes two main modules: feature extraction module and classifier module. In the feature extraction module, a proper set of shape features and statistical feature are proposed as the efficient characteristics of the patterns. In the classifier module, several neural networks, such as multilayer perceptron, probabilistic neural network and radial basis function are investigated. Based on an experimental study, the best classifier is chosen in order to recognize the CCPs. Second, a hybrid heuristic recognition system is introduced based on cuckoo optimization algorithm (COA) algorithm to improve the generalization performance of the classifier. The simulation results show that the proposed algorithm has high recognition accuracy. PMID:24210290

Addeh, Jalil; Ebrahimzadeh, Ata; Azarbad, Milad; Ranaee, Vahid

2014-09-01

304

Suitability of Artificial Neural Network in Daily Flow Forecasting

Directory of Open Access Journals (Sweden)

Full Text Available This study aims to development of the Kasilian indicator river flow forecasting system using Artificial Neural Network (ANN. In this study the performance of multi-layer perceptrons or MLPs, the most frequently used artificial neural network algorithm in the water resources literature, in daily flow estimation and forecasting was investigated. Kasilian watershed in Northern Iran, representing a continuous rain-fall with a predictable stream flow events. Division of yearly data into four seasons and development of separate networks accordingly was found to be more useful than a single network applicable for the entire year. The used data in ANN was hydrometric and climatic daily data with 10 years duration from 1991 to 2000. For the mentioned model 8 years data were used for its development but for the validation/testing of the model 2 years data was applied. Based on the results, the L-M algorithm is more efficient than the CG algorithm, so it is used to train 6 ANNs models for rain fall-runoff prediction at time step t+1 from time step t input. The used network in this study was MLP with BP (back propagation algorithm.

Karim Solaimani

2008-01-01

305

An artificial neural network based matching metric for iris identification

The iris is currently believed to be the most accurate biometric for human identification. The majority of fielded iris identification systems are based on the highly accurate wavelet-based Daugman algorithm. Another promising recognition algorithm by Ives et al uses Directional Energy features to create the iris template. Both algorithms use Hamming distance to compare a new template to a stored database. Hamming distance is an extremely fast computation, but weights all regions of the iris equally. Work from multiple authors has shown that different regions of the iris contain varying levels of discriminatory information. This research evaluates four post-processing similarity metrics for accuracy impacts on the Directional Energy and wavelets based algorithms. Each metric builds on the Hamming distance method in an attempt to use the template information in a more salient manner. A similarity metric extracted from the output stage of a feed-forward multi-layer perceptron artificial neural network demonstrated the most promise. Accuracy tables and ROC curves of tests performed on the publicly available Chinese Academy of Sciences Institute of Automation database show that the neural network based distance achieves greater accuracy than Hamming distance at every operating point, while adding less than one percent computational overhead.

Broussard, Randy P.; Kennell, Lauren R.; Ives, Robert W.; Rakvic, Ryan N.

2008-02-01

306

Yield Stress Modeling of Electrorheological Fluids Using Neural Network

Electrorheological (ER) fluids are a kind of smart materials whose rheological properties can be rapidly changed by applied electric fields. Many potential industrial applications of ER technology have been proposed. In order to formulate better ER fluids and design ER devices, it is important to predict the yield stress of ER fluids based on the ER fluids components and the operating conditions. This paper proposes a new method for predicting the yield stress of ER fluids with neural network (NN). A multilayer perceptron with a single hidden layer of neurons is used to model the ER effect. The data for training and test were produced from the simulation of previous proposed mathematical models. The Levernberg-Marquardt back propagation algorithm was selected for fast learning. The results show the neural network model can well approximate the previous theoretical model, and the predicted outputs of NN agree nearly with the theoretical model values under the same input, all of which demonstrate that it is possible to generate a robust NN model for rapidly predicting the yield stress of ER fluids under different input parameters.

Wei, Kexiang; Meng, Guang

307

An array of commercial gas sensors and nanotechnology sensors has been integrated to quantify gas concentration of air-pollutants. A variety of chemoresistive gas sensors, commercial (Figaro and Fis) and developed at ENEA laboratories (metal-modified carbon nanotubes) were tested to implement a database useful for applied artificial neural networks (ANNs). The ANN algorithm used is the common perceptron multi-layer feed-forward network based on error back-propagation. Electronic Noses based on various sensor arrays related to mammalian olfactory systems have been largely reported [1,2]. Here, we reported on the perceptron-based ANNs applied to a large database of 3875 datapoints for environmental air monitoring. The ANNs performance has been individually assessed for any targeted gas. The response of the classifier has been measured for NO2, CO, CO2, SO2, and H2S gas. The NO2 characteristics exhibit that real concentrations and predicted concentrations are very close with a normalized mean square error (NMSE) in the test set as low as 6%.

Penza, Michele; Suriano, Domenico; Cassano, Gennaro; Rossi, Riccardo; Alvisi, Marco; Pfister, Valerio; Trizio, Livia; Brattoli, Magda; De Gennaro, Gianluigi

2011-09-01

308

Directory of Open Access Journals (Sweden)

Full Text Available The article is devoted to the simulation of the influence of environmental osmotic pressure on the changes of the level of ?-amylase activity of mucous tunic of the intestine of Russian sturgeon (Acipenser güldenstädtii Brandt. For the solving of this problem the apparatus of neural networks is used. The designed model can be classified as multilayer perceptrone and has rather transparent structure. The conformities of this influence are examined and the model with high approximating and generalizing properties is created. The conclusion about high availability of application of the approach in the studies of adaptations of the digestive system of aquatic organisms to the influence of environmental factors with some qualifications about used rate of exactness of the simulation is made.

Bednyakov Dmitriy Andreevich

2012-11-01

309

Artificial neural networks (ANN) have been used in a variety of problems in the fields of science and engineering. Applications of ANN to the geophysical problems have increased within the last decade. In particular, it has been used to solve such inversion problems as seismic, electromagnetic, resistivity. There are also some other applications such as parameter estimation, prediction, and classification. In this study, multilayer perceptron neural networks (MLPNN) and radial basis function neural networks (RBFNN) were applied to synthetic gravity data and Seferihisar gravity data to investigate the applicability and performance of these networks for the method of gravity. Additionally performance of MLPNN and RBFNN were tested by adding random noise to the same synthetic test data. The structure parameters, such as the depths, the density contrasts, and the locations of the structures were obtained closely for different signal-to-noise ratios ( S/ N). Bouguer data of Seferihisar area were analyzed by MLPNN and RBFNN to estimate depth, density contrast, and location of the structure. The results of MLPNN, RBFNN, and classical inversion method were compared for real data obtained from Seferihisar Geothermal area and similar structure parameters were obtained. The experiments show that in general RBFNN not only increases the speed of the training stage enormously, but also provides slightly better performance.

Kaftan, Ilknur; Salk, Mujgan; Senol, Yavuz

2011-12-01

310

Identification and control of plasma vertical position using neural network in Damavand tokamak

International Nuclear Information System (INIS)

In this work, a nonlinear model is introduced to determine the vertical position of the plasma column in Damavand tokamak. Using this model as a simulator, a nonlinear neural network controller has been designed. In the first stage, the electronic drive and sensory circuits of Damavand tokamak are modified. These circuits can control the vertical position of the plasma column inside the vacuum vessel. Since the vertical position of plasma is an unstable parameter, a direct closed loop system identification algorithm is performed. In the second stage, a nonlinear model is identified for plasma vertical position, based on the multilayer perceptron (MLP) neural network (NN) structure. Estimation of simulator parameters has been performed by back-propagation error algorithm using Levenberg–Marquardt gradient descent optimization technique. The model is verified through simulation of the whole closed loop system using both simulator and actual plant in similar conditions. As the final stage, a MLP neural network controller is designed for simulator model. In the last step, online training is performed to tune the controller parameters. Simulation results justify using of the NN controller for the actual plant.

311

Artificial neural network as the tool in prediction rheological features of raw minced meat

Directory of Open Access Journals (Sweden)

Full Text Available Background. The aim of the study was to elaborate a method of modelling and forecasting rheological features which could be applied to raw minced meat at the stage of mixture preparation with a given ingredient composition. Material and methods. The investigated material contained pork and beef meat, pork fat, fat substitutes, ice and curing mixture in various proportions. Seven texture parameters were measured for each sample of raw minced meat. The data obtained were processed using the artificial neural network module in Statistica 9.0 software. Results. The model that reached the lowest training error was a multi-layer perceptron MLP with three neural layers and architecture 7:7-11-7:7. Correlation coefficients between the experimental and calculated values in training, verification and testing subsets were similar and rather high (around 0.65 which indicated good network performance. Conclusion. High percentage of the total variance explained in PCA analysis (73.5% indicated that the percentage composition of raw minced meat can be successfully used in the prediction of its rheological features. Statistical analysis of the results revealed, that artificial neural network model is able to predict rheological parameters and thus a complete texture profile of raw minced meat.

Edyta Balejko

2012-09-01

312

Modeling soil temperatures at different depths by using three different neural computing techniques

This study compares the accuracy of three different neural computing techniques, multi-layer perceptron (MLP), radial basis neural networks (RBNN), and generalized regression neural networks (GRNN), in modeling soil temperatures (ST) at different depths. Climatic data of air temperature, wind speed, solar radiation, and relative humidity from Mersin Station, Turkey, were used as inputs to the models to estimate monthly ST values. In the first part of the study, the effect of each climatic variable on ST was investigated by using GRNN models. Air temperature was found to be the most effective variable in modeling monthly ST. In the second part of the study, the accuracy of GRNN models was compared with MLP, RBNN, and multiple linear regression (MLR) models. RBNN models were found to be better than the GRNN, MLP, and MLR models in estimating monthly ST at the depths of 5 and 10 cm while the MLR and GRNN models gave the best accuracy in the case of 50- and 100-cm depths, respectively. In the third part of the study, the effect of periodicity on the training, validation, and test accuracy of the applied models was investigated. The results indicated that the adding periodicity component significantly increase models' accuracies in estimating monthly ST at different depths. Root mean square errors of the GRNN, MLP, RBNN, and MLR models were decreased by 19, 15, 19, and 15 % using periodicity in estimating monthly ST at 5-cm depth.

Kisi, Ozgur; Tombul, Mustafa; Kermani, Mohammad Zounemat

2014-08-01

313

A new approach to self-organizing fuzzy polynomial neural networks guided by genetic optimization

In this study, we introduce a new topology of Fuzzy Polynomial Neural Networks (FPNN) that is based on a genetically optimized multilayer perceptron with fuzzy polynomial neurons (FPNs) and discuss its comprehensive design methodology. The underlying methodology involves mechanisms of genetic optimization, especially genetic algorithms (GAs). Let us recall that the design of the “conventional” FPNNs uses an extended Group Method of Data Handling (GMDH) and exploits a fixed fuzzy inference type located at each FPN of the FPNN as well as considers a fixed number of input nodes at FPNs (or nodes) located in each layer. The proposed FPNN gives rise to a structurally optimized structure and comes with a substantial level of flexibility in comparison to the one we encounter in conventional FPNNs. The structural optimization is realized via GAs whereas in the case of the parametric optimization we proceed with a standard least square method based learning. Through the consecutive process of such structural and parametric optimization, an optimized and flexible fuzzy neural network is generated in a dynamic fashion. The performance of the proposed gFPNN is quantified through experimentation that exploits standard data already being used in fuzzy modeling. The results reveal superiority of the proposed networks over the existing fuzzy and neural models.

Oh, Sung-Kwun; Pedrycz, Witold

2005-09-01

314

Neural networks forecast in small catchments with transfer of network parameters

This contribution deals with neural network approach for short term forecast on small catchments. The applied methodology is based on theory of multilayer perceptron (MLP), feed forward neural network with back propagation optimization procedure was tested in order to explore the possibilities to transfer parameters between different catchments. Supervised optimization of network parameters and structure was investigated. A software tool was created for these research and operative purposes. The hourly discharges and rainfall data of real flood events served as an input to MLP. Seven catchments with areas, which range from 10 to 250 square kilometres and which are situated in the east part of the Czech Republic, were selected. The input data were normalized by parametric method. Variable configuration of neural network was tested in number of modes represented by different combination of learning and testing data sets. The analysis focuses on ability of the model to forecast the flood event with different peak discharge magnitudes. This should be achieved in both application steps - MLP learning and testing within given catchment and in step of parameter transfer of well learned network to another catchment. The length of prediction ranged from one hour to six hours ahead. The results showed that the model is capable to provide satisfying short term discharge forecast for the most of studied cases, including successful parameter transfer among different catchments. This was accomplished by using optimization of parameters which determine not only the structure and behaviour of applied network but also the transformation of input data.

Maca, P.; Havlicek, V.; Hermanovsky, M.; Horacek, S.; Pech, P.

2009-04-01

315

Artificial neural networks (ANNs) were applied for system understanding and prediction of drug release properties from direct compacted matrix tablets using sucrose esters (SEs) as matrix-forming agents for controlled release of a highly water soluble drug, metoprolol tartrate. Complexity of the system was presented through the effects of SE concentration and tablet porosity at various hydrophilic-lipophilic balance (HLB) values of SEs ranging from 0 to 16. Both effects contributed to release behaviors especially in the system containing hydrophilic SEs where swelling phenomena occurred. A self-organizing map neural network (SOM) was applied for visualizing interrelation among the variables and multilayer perceptron neural networks (MLPs) were employed to generalize the system and predict the drug release properties based on HLB value and concentration of SEs and tablet properties, i.e., tablet porosity, volume and tensile strength. Accurate prediction was obtained after systematically optimizing network performance based on learning algorithm of MLP. Drug release was mainly attributed to the effects of SEs, tablet volume and tensile strength in multi-dimensional interrelation whereas tablet porosity gave a small impact. Ability of system generalization and accurate prediction of the drug release properties proves the validity of SOM and MLPs for the formulation modeling of direct compacted matrix tablets containing controlled release agents of different material properties. PMID:21878388

Chansanroj, Krisanin; Petrovi?, Jelena; Ibri?, Svetlana; Betz, Gabriele

2011-10-01

316

Identification and control of plasma vertical position using neural network in Damavand tokamak

In this work, a nonlinear model is introduced to determine the vertical position of the plasma column in Damavand tokamak. Using this model as a simulator, a nonlinear neural network controller has been designed. In the first stage, the electronic drive and sensory circuits of Damavand tokamak are modified. These circuits can control the vertical position of the plasma column inside the vacuum vessel. Since the vertical position of plasma is an unstable parameter, a direct closed loop system identification algorithm is performed. In the second stage, a nonlinear model is identified for plasma vertical position, based on the multilayer perceptron (MLP) neural network (NN) structure. Estimation of simulator parameters has been performed by back-propagation error algorithm using Levenberg-Marquardt gradient descent optimization technique. The model is verified through simulation of the whole closed loop system using both simulator and actual plant in similar conditions. As the final stage, a MLP neural network controller is designed for simulator model. In the last step, online training is performed to tune the controller parameters. Simulation results justify using of the NN controller for the actual plant.

Rasouli, H.; Rasouli, C.; Koohi, A.

2013-02-01

317

Identification and control of plasma vertical position using neural network in Damavand tokamak

Energy Technology Data Exchange (ETDEWEB)

In this work, a nonlinear model is introduced to determine the vertical position of the plasma column in Damavand tokamak. Using this model as a simulator, a nonlinear neural network controller has been designed. In the first stage, the electronic drive and sensory circuits of Damavand tokamak are modified. These circuits can control the vertical position of the plasma column inside the vacuum vessel. Since the vertical position of plasma is an unstable parameter, a direct closed loop system identification algorithm is performed. In the second stage, a nonlinear model is identified for plasma vertical position, based on the multilayer perceptron (MLP) neural network (NN) structure. Estimation of simulator parameters has been performed by back-propagation error algorithm using Levenberg-Marquardt gradient descent optimization technique. The model is verified through simulation of the whole closed loop system using both simulator and actual plant in similar conditions. As the final stage, a MLP neural network controller is designed for simulator model. In the last step, online training is performed to tune the controller parameters. Simulation results justify using of the NN controller for the actual plant.

Rasouli, H. [School of Plasma Physics and Nuclear Fusion, Institute of Nuclear Science and Technology, AEOI, P.O. Box 14155-1339, Tehran (Iran, Islamic Republic of); Advanced Process Automation and Control (APAC) Research Group, Faculty of Electrical Engineering, K.N. Toosi University of Technology, P.O. Box 16315-1355, Tehran (Iran, Islamic Republic of); Rasouli, C.; Koohi, A. [School of Plasma Physics and Nuclear Fusion, Institute of Nuclear Science and Technology, AEOI, P.O. Box 14155-1339, Tehran (Iran, Islamic Republic of)

2013-02-15

318

Selection of input parameters to model direct solar irradiance by using artificial neural networks

International Nuclear Information System (INIS)

A very important factor in the assessment of solar energy resources is the availability of direct irradiance data of high quality. However, this component of solar radiation is seldom measured and thus must be estimated from data of global solar irradiance, which is registered in most radiometric stations. In recent years, artificial neural networks (ANN) have shown to be a powerful tool for mapping complex and non-linear relationships. In this work, the Bayesian framework for ANN, named as automatic relevance determination method (ARD), was employed to obtain the relative relevance of a large set of atmospheric and radiometric variables used for estimating hourly direct solar irradiance. In addition, we analysed the viability of this novel technique applied to select the optimum input parameters to the neural network. For that, a multi-layer feedforward perceptron is trained on these data. The results reflect the relative importance of the inputs selected. Clearness index and relative air mass were found to be the more relevant input variables to the neural network, as it was expected, proving the reliability of the ARD method. Moreover, we show that this novel methodology can be used in unfavourable conditions, in terms of limited amount of available data, performing successful results

319

The Role of Weight Shrinking in Large Margin Perceptron Learning

We introduce into the classical perceptron algorithm with margin a mechanism that shrinks the current weight vector as a first step of the update. If the shrinking factor is constant the resulting algorithm may be regarded as a margin-error-driven version of NORMA with constant learning rate. In this case we show that the allowed strength of shrinking depends on the value of the maximum margin. We also consider variable shrinking factors for which there is no such dependence. In both cases we obtain new generalizations of the perceptron with margin able to provably attain in a finite number of steps any desirable approximation of the maximal margin hyperplane. The new approximate maximum margin classifiers appear experimentally to be very competitive in 2-norm soft margin tasks involving linear kernels.

Panagiotakopoulos, Constantinos

2012-01-01

320

Deep Big Simple Neural Nets Excel on Handwritten Digit Recognition

Good old on-line back-propagation for plain multi-layer perceptrons yields a very low 0.35% error rate on the famous MNIST handwritten digits benchmark. All we need to achieve this best result so far are many hidden layers, many neurons per layer, numerous deformed training images, and graphics cards to greatly speed up learning.

Ciresan, Dan Claudiu; Gambardella, Luca Maria; Schmidhuber, Juergen

2010-01-01

321

Tagging b quark events in ALEPH with neural networks

International Nuclear Information System (INIS)

Comparison of different methods to tag b quark events are presented: multilayered perceptron (MLP), Learning Vector Quantization (LVQ), discriminant analysis, combination of any two of the above methods. The sample events come from the ALEPH Monte Carlo and data, from the 1990 ALEPH runs. (authors) 12 refs., 16 figs., 5 tabs

322

ARTMAP and orthonormal basis function neural networks for pattern classification

This dissertation investigates neural network approaches to pattern classification. One application considered is the classification of land use change in the Nile River delta between 1984 and 1993 from ten Landsat Thematic Mapper (Landsat TM) images acquired during this period. Other applications, including image segmentation, letter recognition, and prediction of variables from census data, are represented by the standardized DELVE (Data for Evaluating Learning in Valid Experiments) machine learning database. An ARTMAP (Adaptive Resonance Theory Map) neural network system is developed for the land use change classification task. Cross-validation is used to enable design decisions and to enable model fitting to be done without regard to data in test partitions. The training of voting ARTMAP systems on brightness-greenness-wetness (BGW) data for multiple dates and location data results in performance competitive with previously used expert systems. Orthonormal basis function classification methods are extended to make them appropriate for multidimensional problems. These methods share the multilayer perceptron architecture common to many neural networks. A layer of basis functions transforms the data prior to classification. Stopping rules are used to determine which basis functions to include in a model to minimize the expected mean integrated squared error (MISE). To perform stopping when using the discriminant function of Devroye et al. (1996), an appropriate MISE estimator is developed. Linear transformations to rotate data and improve multiple classification results are investigated using development benchmarks from the DELVE suite. Orthonormal basis function neural network classifiers using these principles are developed and tested along with standard pattern classification techniques on the DELVE suite. Orthonormal basis function systems appear to be well suited for some multidimensional problems. These systems, along with benchmark classifiers, are also applied to the Nile River delta dataset. Although orthonormal basis function systems are an appropriate choice for this task, the best performance observed on this dataset is that of linear discriminant analysis (LDA) applied to multitemporal data.

Shock, Byron Mitchell

323

Data acquisition in modeling using neural networks and decision trees

Directory of Open Access Journals (Sweden)

Full Text Available The paper presents a comparison of selected models from area of artificial neural networks and decision trees in relation with actualconditions of foundry processes. The work contains short descriptions of used algorithms, their destination and method of data preparation,which is a domain of work of Data Mining systems. First part concerns data acquisition realized in selected iron foundry, indicating problems to solve in aspect of casting process modeling. Second part is a comparison of selected algorithms: a decision tree and artificial neural network, that is CART (Classification And Regression Trees and BP (Backpropagation in MLP (Multilayer Perceptron networks algorithms.Aim of the paper is to show an aspect of selecting data for modeling, cleaning it and reducing, for example due to too strong correlationbetween some of recorded process parameters. Also, it has been shown what results can be obtained using two different approaches:first when modeling using available commercial software, for example Statistica, second when modeling step by step using Excel spreadsheetbasing on the same algorithm, like BP-MLP. Discrepancy of results obtained from these two approaches originates from a priorimade assumptions. Mentioned earlier Statistica universal software package, when used without awareness of relations of technologicalparameters, i.e. without user having experience in foundry and without scheduling ranks of particular parameters basing on acquisition, can not give credible basis to predict the quality of the castings. Also, a decisive influence of data acquisition method has been clearly indicated, the acquisition should be conducted according to repetitive measurement and control procedures. This paper is based on about 250 records of actual data, for one assortment for 6 month period, where only 12 data sets were complete (including two that were used for validation of neural network and useful for creating a model. It is definitely too small portion in case of artificial neural networks, but it shows a scale of danger of unprofessional data acquisition.

R. Sika

2011-04-01

324

Artificial Neural Networks in the Assessment of Stand Parameters from an IKONOS Satellite Image

Directory of Open Access Journals (Sweden)

Full Text Available The paper explores the possibilities of assessing five stand parameters (tree number, volume, stocking, basal area and stand age with the application of a multi-layer perceptron artificial neural network. An IKONOS satellite image (PAN 1 m x 1 m was used to asses parts of stands in the sixth (121–140 yrs and seventh (141–160 yrs age class of pedunculate oak management class in the »Slavir« Management Unit of Otok Forest Office. Six features extracted from the first order histogram and five texture features extracted from the second order histogram were used as input data for neural network training. Data from the Management Plan were used as outputs of the neural network. An early stopping method and scaled conjugate gradient algorithm with error back propagation were used to improve generalization property of the neural network. Two neural network models were applied to assess the required stand parameters. The first model has one neuron in the output layer, where separate neuron network training was conducted for each stand parameter. The second model has five neurons in the output layer related to five assessed stand parameters. Both networks were trained and tested simultaneously. The conducted research showed that both of these neuron network models have good generalization properties. However, further analysis gave precedence to the second neural network model. Assessment of five quantitative stand parameters did not show any statistically significant differences between the Management Plan data and the neuron network model in terms of tree number, volume, stocking, basal area and stand age analysis.

Damir Klobu?ar

2008-12-01

325

This paper presents a comparative analysis of novel supervised fuzzy adaptive resonance theory (SF-ART), multilayer perceptron (MLP) and Multi Layer Perceptrons (MLP) neural networks over Ballistocardiogram (BCG) signal recognition. To extract essential features of the BCG signal, we applied Biorthogonal wavelets. SF-ART performs classification on two levels. At first level, pre-classifier which is self-organized fuzzy ART tuned for fast learning classifies the input data roughly to arbitrary (M) classes. At the second level, post-classification level, a special array called Affine Look-up Table (ALT) with M elements stores the labels of corresponding input samples in the address equal to the index of fuzzy ART winner. However, in running (testing) mode, the content of an ALT cell with address equal to the index of fuzzy ART winner output will be read. The read value declares the final class that input data belongs to. In this paper, we used two well-known patterns (IRIS and Vowel data) and a medical application (Ballistocardiogram data) to evaluate and check SF-ART stability, reliability, learning speed and computational load. Initial tests with BCG from six subjects (both healthy and unhealthy people) indicate that the SF-ART is capable to perform with a high classification performance, high learning speed (elapsed time for learning around half second), and very low computational load compared to the well-known neural networks such as MLP which needs minutes to learn the training material. Moreover, to extract essential features of the BCG signal, we applied Biorthogonal wavelets. The applied wavelet transform requires no prior knowledge of the statistical distribution of data samples. PMID:17283924

Akhbardeh, Alireza; Junnila, Sakari; Koivistoinen, Teemu; Värri, Alpo

2007-02-01

326

Crash Introduction to Artificial Neural Networks

The Crash Introduction to Artificial Neural Networks is not a comprehensive resource, but it provides a good overview of many aspects of the topic. Beginning with a discussion of biological processes in the brain, the site describes the function of neurons and how they are interconnected. Some historical events are mentioned, leading to the development of an artificial neural network. The famous perceptron configuration is the basis of subsequent discussions of training algorithms, prediction and classification functions, and data processing.

Galkin, Ivan

327

Directory of Open Access Journals (Sweden)

Full Text Available Direct measurement of soil hydraulic characteristics is costly and time-consuming. Also, the method is partly unreliable due to soil heterogeneity and laboratory errors. Instead, soil hydraulic characteristics can be predicted using readily available data such as soil texture and bulk density using pedotransfer functions (PTFs. Artificial neural networks (ANNs and statistical regression are two methods which are used to develop PTFs. In this study, the multi-layer perceptron (MLP neural network and backward and stepwise regression models were used to estimate saturated hydraulic conductivity using some soil characteristics including the percentage of particle size distribution, porosity, and bulk density. Data of 125 soil profiles were collected from the reports of basic soil science and land reclamation studies conducted by Khuzestan Water and Power Organization. The results showed that MLP neural network having Bayesian training algorithm with the greater coefficient of determination (R2=0.65 and the lower error (RMSE =0.04 had better performance than multiple linear regression model in predicting saturated hydraulic conductivity.

R. Rezae Arshad

2012-07-01

328

In computational physics proton transfer phenomena could be viewed as pattern classification problems based on a set of input features allowing classification of the proton motion into two categories: transfer ‘occurred’ and transfer ‘not occurred’. The goal of this paper is to evaluate the use of artificial neural networks in the classification of proton transfer events, based on the feed-forward back propagation neural network, used as a classifier to distinguish between the two transfer cases. In this paper, we use a new developed data mining and pattern recognition tool for automating, controlling, and drawing charts of the output data of an Empirical Valence Bond existing code. The study analyzes the need for pattern recognition in aqueous proton transfer processes and how the learning approach in error back propagation (multilayer perceptron algorithms) could be satisfactorily employed in the present case. We present a tool for pattern recognition and validate the code including a real physical case study. The results of applying the artificial neural networks methodology to crowd patterns based upon selected physical properties (e.g., temperature, density) show the abilities of the network to learn proton transfer patterns corresponding to properties of the aqueous environments, which is in turn proved to be fully compatible with previous proton transfer studies.

Amani, Tahat; Jordi, Marti; Ali, Khwaldeh; Kaher, Tahat

2014-04-01

329

Energy Technology Data Exchange (ETDEWEB)

Global solar radiation (GSR) data are desirable for many areas of research and applications in various engineering fields. However, GSR is not as readily available as air temperature data. Artificial neural networks (ANNs) are effective tools to model nonlinear systems and require fewer inputs. The objective of this study was to test an artificial neural network (ANN) for estimating the global solar radiation (GSR) as a function of air temperature data in a semi-arid environment. The ANNs (multilayer perceptron type) were trained to estimate GSR as a function of the maximum and minimum air temperature and extraterrestrial radiation. The data used in the network training were obtained from a historical series (1994-2001) of daily climatic data collected in weather station of Ahwaz located in Khuzestan plain in the southwest of Iran. The empirical Hargreaves and Samani equation (HS) is also considered for the comparison. The HS equation calibrated by applying the same data used for neural network training. Two historical series (2002-2003) were utilized to test the network and for comparison between the ANN and calibrated HS method. The study demonstrated that modelling of daily GSR through the use of the ANN technique gave better estimates than the HS equation. RMSE and R{sup 2} for the comparison between observed and estimated GSR for the tested data using the proposed ANN model are 2.534 MJ m{sup -2} day{sup -1} and 0.889 respectively. (author)

Rahimikhoob, Ali [Irrigation and Drainage Engineering Department, College of Abouraihan, University of Tehran (Iran)

2010-09-15

330

Perceptron capacity revisited: classification ability for correlated patterns

Energy Technology Data Exchange (ETDEWEB)

In this paper, we address the problem of how many randomly labeled patterns can be correctly classified by a single-layer perceptron when the patterns are correlated with each other. In order to solve this problem, two analytical schemes are developed based on the replica method and the Thouless-Anderson-Palmer (TAP) approach by utilizing an integral formula concerning random rectangular matrices. The validity and relevance of the developed methodologies are shown for one known result and two example problems. A message-passing algorithm to perform the TAP scheme is also presented.

Shinzato, Takashi; Kabashima, Yoshiyuki [Department of Computational Intelligence and Systems Science, Tokyo Institute of Technology, Yokohama 226-8502 (Japan)], E-mail: shinzato@sp.dis.titech.ac.jp, E-mail: kaba@dis.titech.ac.jp

2008-08-15

331

Learning algorithm that gives the Bayes generalization limit for perceptrons

A variational approach to the study of learning a linearly separable rule by a single-layer perceptron leads to a gradient descent learning algorithm with exactly the same generalization ability as the Bayes limit calculated by Opper and Haussler [Phys. Rev. Lett. 66, 2677 (1991)]. This is done by finding, through the Gardner-Derrida replica method, the student-teacher overlap R as a functional of the algorithm cost function and maximizing this functional. The resulting cost function is closely related to the optimal cost function derived for on-line learning.

Kinouchi, Osame; Caticha, Nestor

1996-07-01

332

Perceptron capacity revisited: classification ability for correlated patterns

In this paper, we address the problem of how many randomly labeled patterns can be correctly classified by a single-layer perceptron when the patterns are correlated with each other. In order to solve this problem, two analytical schemes are developed based on the replica method and the Thouless-Anderson-Palmer (TAP) approach by utilizing an integral formula concerning random rectangular matrices. The validity and relevance of the developed methodologies are shown for one known result and two example problems. A message-passing algorithm to perform the TAP scheme is also presented.

Shinzato, Takashi; Kabashima, Yoshiyuki

2008-08-01

333

Alternative target functions for protein structure prediction with neural networks

The prediction and modeling of protein structure is a central problem in bioinformatics. Neural networks have been used extensively to predict the secondary structure of proteins. While significant progress has been made by using multiple sequence data, the ability to predict secondary structure from a single sequence and a single prediction network has stagnated with an accuracy of about 75%. This implies that there is some limit to the accuracy of the prediction. In order to understand this behavior we asked the question of what happens as we change the target function for the prediction. Instead of predicting a derived quantity, such as whether a given chain is a helix, sheet or turn, we tested whether a more directly observed quantity such as the distance between a pair of ?-carbon atoms could be predicted with reasonable accuracy. The ?-carbon atom position is central to each residue in the protein and the distances between them in sequence define the backbone of protein. Knowledge of the distances between the ?-carbon atoms is sufficient to determine the three dimensional structure of the protein. We have trained on distance data derived from the complete protein structure database (pdb) using a multi-layered perceptron feedforward neural network with back propagation. It shows that the root of mean square error is 0.4 Å with orthogonal coding of protein primary sequence. This is comparable to the experimental error in the structures used to form the database. The effects of exploring other encoding schemes, and different complexities of neural networks as well as related target functions such as distance thresholds will be presented.

Deng, Hai; Harrison, Robert; Pan, Yi; Tai, Phang C.

2004-04-01

334

The reactor safety study with help of artificial neuron networks (multilayer perceptrons)

International Nuclear Information System (INIS)

One deals with deposition of insulation large amounts on settling tank components that may result in malfunction of residual heat removal systems. Paper describes briefly simulation of pressure drops in confinement systems by means of an artificial neuron nets and compares the simulation data with the experiment ones

335

Classification of Polar-Thermal Eigenfaces using Multilayer Perceptron for Human Face Recognition

Digital Repository Infrastructure Vision for European Research (DRIVER)

This paper presents a novel approach to handle the challenges of face recognition. In this work thermal face images are considered, which minimizes the affect of illumination changes and occlusion due to moustache, beards, adornments etc. The proposed approach registers the training and testing thermal face images in polar coordinate, which is capable to handle complicacies introduced by scaling and rotation. Polar images are projected into eigenspace and finally classified ...

Bhowmik, Mrinal Kanti; Bhattacharjee, Debotosh; Nasipuri, Mita; Basu, Dipak Kumar; Kundu, Mahantapas

2010-01-01

336

Most real and engineered systems include multiple subsystems and layers of connectivity, and it is important to take such features into account to try to obtain a complete understanding of these systems. It is thus necessary to generalize "traditional" network theory by developing (and validating) a framework and associated tools to study multilayer systems in a comprehensive fashion. The origins of such efforts occurred several decades ago, but now the study of multilayer networks has become one of the major directions in network science. In this paper, we discuss the history of multilayer networks (and related concepts) and then review the exploding body of work on such networks. To unify the disparate terminology in the large body of recent work, we discuss a general framework for multilayer networks, construct a dictionary of terminology to relate the numerous existing concepts to each other, and provide a thorough discussion that compares, contrasts, and translates between related notions such as multila...

Kivelä, Mikko; Barthelemy, Marc; Gleeson, James P; Moreno, Yamir; Porter, Mason A

2013-01-01

337

Target discrimination in synthetic aperture radar using artificial neural networks.

This paper addresses target discrimination in synthetic aperture radar (SAR) imagery using linear and nonlinear adaptive networks. Neural networks are extensively used for pattern classification but here the goal is discrimination. We show that the two applications require different cost functions. We start by analyzing with a pattern recognition perspective the two-parameter constant false alarm rate (CFAR) detector which is widely utilized as a target detector in SAR. Then we generalize its principle to construct the quadratic gamma discriminator (QGD), a nonparametrically trained classifier based on local image intensity. The linear processing element of the QCD is further extended with nonlinearities yielding a multilayer perceptron (MLP) which we call the NL-QGD (nonlinear QGD). MLPs are normally trained based on the L(2) norm. We experimentally show that the L(2) norm is not recommended to train MLPs for discriminating targets in SAR. Inspired by the Neyman-Pearson criterion, we create a cost function based on a mixed norm to weight the false alarms and the missed detections differently. Mixed norms can easily be incorporated into the backpropagation algorithm, and lead to better performance. Several other norms (L(8), cross-entropy) are applied to train the NL-QGD and all outperformed the L(2) norm when validated by receiver operating characteristics (ROC) curves. The data sets are constructed from TABILS 24 ISAR targets embedded in 7 km(2) of SAR imagery (MIT/LL mission 90). PMID:18276330

Principe, J C; Kim, M; Fisher, M

1998-01-01

338

Identification and Prediction of Internet Traffic Using Artificial Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available This paper presents the development of an artificial neural network (ANN model based on the multi-layer perceptron (MLP for analyzing internet traffic data over IP networks. We applied the ANN to analyze a time series of measured data for network response evaluation. For this reason, we used the input and output data of an internet traffic over IP networks to identify the ANN model, and we studied the performance of some training algorithms used to estimate the weights of the neuron. The comparison between some training algorithms demonstrates the efficiency and the accu-racy of the Levenberg-Marquardt (LM and the Resilient back propagation (Rp algorithms in term of statistical crite-ria. Consequently, the obtained results show that the developed models, using the LM and the Rp algorithms, can successfully be used for analyzing internet traffic over IP networks, and can be applied as an excellent and fundamental tool for the management of the internet traffic at different times.

Samira Chabaa

2010-09-01

339

AN EFFICIENT NEURAL NETWORK FOR RECOGNIZING GESTURAL HINDI DIGITS

Directory of Open Access Journals (Sweden)

Full Text Available Handwritten Hindi digit recognition plays an important role in eastern Arab countries especially in the courtesy amounts of Arab bank checks, recognizing numbers in car plates, or in postal code for mail sorting. In our study, we proposed an efficient Hindi Digit Recognition System drawn by the mouse and developed using Multilayer Perceptron Neural Network (MLP with backpropagation. The system has been designed, implemented and tested successfully. Analysis has been carried out to determine the number of hidden nodes that achieves high performance. The proposed system has been trained on samples of 800 images and tested on samples of 300 images written by different users selected from different ages. An experimental result shows high accuracy of about 91% on the testing samples and very close to 100% on the training samples. Experiments showed that our result is high in comparison with other Hindi digit recognition systems especially if we consider the way of writing (mouse and children in our trained and tested results.

Nidal Fawzi Shilbayeh

2013-01-01

340

Monthly monsoon rainfall forecasting using artificial neural networks

Indian agriculture sector heavily depends on monsoon rainfall for successful harvesting. In the past, prediction of rainfall was mainly performed using regression models, which provide reasonable accuracy in the modelling and forecasting of complex physical systems. Recently, Artificial Neural Networks (ANNs) have been proposed as efficient tools for modelling and forecasting. A feed-forward multi-layer perceptron type of ANN architecture trained using the popular back-propagation algorithm was employed in this study. Other techniques investigated for modeling monthly monsoon rainfall include linear and non-linear regression models for comparison purposes. The data employed in this study include monthly rainfall and monthly average of the daily maximum temperature in the North Central region in India. Specifically, four regression models and two ANN model's were developed. The performance of various models was evaluated using a wide variety of standard statistical parameters and scatter plots. The results obtained in this study for forecasting monsoon rainfalls using ANNs have been encouraging. India's economy and agricultural activities can be effectively managed with the help of the availability of the accurate monsoon rainfall forecasts.

Ganti, Ravikumar

2014-10-01

341

Energy demand estimation of South Korea using artificial neural network

International Nuclear Information System (INIS)

Because South Korea's industries depend heavily on imported energy sources (fifth largest importer of oil and second largest importer of liquefied natural gas in the world), the accurate estimating of its energy demand is critical in energy policy-making. This research proposes an artificial neural network model (a structure with feed-forward multilayer perceptron, error back-propagation algorithm, momentum process, and scaled data) to efficiently estimate the energy demand for South Korea. The model has four independent variables, such as gross domestic product (GDP), population, import, and export amounts. The data are obtained from diverse local and international sources. The proposed model better estimated energy demand than a linear regression model (a structure with multiple linear variables and least square method) or an exponential model (a structure with mixed integer variables, branch and bound method, and Broyden-Fletcher-Goldfarb-Shanno (BFGS) method) in terms of root mean squared error (RMSE). The model also forecasted better than the other two models in terms of RMSE without any over-fitting problem. Further testing with four scenarios based upon reliable source data showed unanticipated results. Instead of growing permanently, the energy demands peaked at certain points, and then decreased gradually. This trend is quite different from the results by regression or exponential model.

342

Energy demand estimation of South Korea using artificial neural network

Energy Technology Data Exchange (ETDEWEB)

Because South Korea's industries depend heavily on imported energy sources (fifth largest importer of oil and second largest importer of liquefied natural gas in the world), the accurate estimating of its energy demand is critical in energy policy-making. This research proposes an artificial neural network model (a structure with feed-forward multilayer perceptron, error back-propagation algorithm, momentum process, and scaled data) to efficiently estimate the energy demand for South Korea. The model has four independent variables, such as gross domestic product (GDP), population, import, and export amounts. The data are obtained from diverse local and international sources. The proposed model better estimated energy demand than a linear regression model (a structure with multiple linear variables and least square method) or an exponential model (a structure with mixed integer variables, branch and bound method, and Broyden-Fletcher-Goldfarb-Shanno (BFGS) method) in terms of root mean squared error (RMSE). The model also forecasted better than the other two models in terms of RMSE without any over-fitting problem. Further testing with four scenarios based upon reliable source data showed unanticipated results. Instead of growing permanently, the energy demands peaked at certain points, and then decreased gradually. This trend is quite different from the results by regression or exponential model. (author)

Geem, Zong Woo [Environmental Planning and Management Program, Johns Hopkins University, Clarksburg, Maryland 20871 (United States); Roper, William E. [Department of Geography and Geo-information Science, George Mason University, Fairfax, Virginia 22030 (United States)

2009-10-15

343

Training feed-forward neural networks using conjugate gradients

Neural networks for optical character recognition are still being trained using back propagation, even though conjugate gradient methods have been shown to be much faster. Most multilayer perceptron network training results in the literature are obtained for small and unrealistic problems or from data sets that are proprietary and not available for comparison testing. We present results on a large realistic pattern set containing 2000 training and 1434 testing exemplars. Each pattern is composed of 32 Gabor coefficients obtained from a 32 by 32 pixel binary image of a handwritten digit segmented from the NIST Handwriting Image Data Base. These sets are believed to have approximately 1 segmentation errors. Comparative results for Moller''s scaled conjugate gradient method and for standard back propagation are presented for runs on a serial scientific workstation and a highly parallel computer. Typical training on a network with 32 inputs, 32 hidden nodes, and 10 output nodes gives a 98 recognition for the training set and 95 for the test set. Training with conjugate gradients requires fewer than 200 iterations; times are about 20 to 40 minutes on a scientific workstation and 6 minutes on the highly parallel computer. Testing (classification) is done at the rate of 600 to 1600 patterns per second on the scientific workstation and on the highly parallel computer respectively. These results suggest that commercial handwritten character recognition systems with great economic potential are feasible.

Blue, James L.; Grother, Patrick J.

1992-08-01

344

Entropy landscape of solutions in the binary perceptron problem

International Nuclear Information System (INIS)

The statistical picture of the solution space for a binary perceptron is studied. The binary perceptron learns a random classification of input random patterns by a set of binary synaptic weights. The learning of this network is difficult especially when the pattern (constraint) density is close to the capacity, which is supposed to be intimately related to the structure of the solution space. The geometrical organization is elucidated by the entropy landscape from a reference configuration and of solution-pairs separated by a given Hamming distance in the solution space. We evaluate the entropy at the annealed level as well as replica symmetric level and the mean field result is confirmed by the numerical simulations on single instances using the proposed message passing algorithms. From the first landscape (a random configuration as a reference), we see clearly how the solution space shrinks as more constraints are added. From the second landscape of solution-pairs, we deduce the coexistence of clustering and freezing in the solution space. (paper)

345

Creep Crack Growth Modeling of Low Alloy Steel using Artificial Neural Network

Directory of Open Access Journals (Sweden)

Full Text Available Prediction of crack growth under creep condition is prime requirement in order to avoid costly and time-consuming creep crack growth tests. To predict, in a reliable way, the growth of a major crack in a structural components operating at high temperatures, requires a fracture mechanics based approach. In this Study a novel technique, which uses Finite Element Method (FEM together with Artificial Neural Networks (ANN has been developed to predict the fracture mechanics parameter (C* in a 1%Cr1%MoV low alloy rotor steel under wide range of loading and temperatures. After confirming the validity of the FEM model with experimental data, a collection of numerical and experimental data has been used for training the various neural networks models. Three networks have been used to simulate the process, the perceptron multilayer network with tangent transfer function that uses 9 neurons in the hidden layer, gives the best results. Finally, for validation three case studies at 538°C, 550°C and 594°C temperatures are employed. The proposed model has proved that a combinations of ANN and FEM simulation performs well in estimation of C* and it is a powerful designing tool for creep crack growth characterization.

F. Djavanroodi

2013-07-01

346

Effect of Heat Fluxes on Ammonia Emission from Swine Waste Lagoon Based on Neural Network Analyses

Directory of Open Access Journals (Sweden)

Full Text Available Understanding factors that affect ammonia emissions from swine waste lagoons or any animal waste receptacles is a necessary first step in deploying potential remediation options. In this study, we examined the various meteorological factors (i.e., air temperatures, solar radiation and heat fluxes that potentially affect ammonia emissions from swine waste lagoon. Ammonia concentrations were monitored using a photoacoustic gas analyzer. The ammonia emissions from the lagoon were monitored continuously for a 24 h cycle, twice a week during a winter month at a height of 50 cm above the lagoon surface. Meteorological data were also monitored simultaneously. Heat fluxes were tabulated and correlated to the averaged ammonia concentrations (range of zero to 8.0 ppmv. Multi-layer Perceptron (MLP neural network predictive model was built based on the most important meteorological parameters. The results from MLP neural networks analysis show that ammonia emissions from the swine waste lagoon were affected by heat fluxes such as net solar radiation, sensible heat and latent heat of vaporization. Thus it is important to consider environmental conditions (i.e., meteorological parameters such as solar radiation, latent heat and etc. in formulating management or abatement strategies for reducing ammonia emissions from swine waste lagoons or any other air pollutant emissions from livestock waste receptacles.

N. Lovanh

2014-01-01

347

THYROID DISEASE DETECTION USING MODIFIED FUZZY HYPERLINE SEGMENT CLUSTERING NEURAL NETWORK

Directory of Open Access Journals (Sweden)

Full Text Available Two common diseases of the thyroid gland, which releases thyroid hormones for regulating the rate of body’s metabolism, are hyperthyroidism and hypothyroidism. The classification of these thyroid diseases is a one of the considerable tasks. In this work we propose modified fuzzy hyperline segment clustering neural network (MFHLSCNN for classification of thyroid disease diagnosis. The MFHLSCNN algorithm is suitable for clustering and classification. This algorithm can learn ill-defined nonlinear cluster boundaries in a few passes and is suitable for on-line adaptation. The work is extension of fuzzy hyperline segment clustering neural network (FHLSCNN proposed by Kulkarni U. V. and Sontakke T. R. Both the algorithms utilize fuzzy sets as pattern clusters in which each fuzzy set is an union of fuzzy set hyperline segments. The fuzzy set hyperline segment is an n-dimensional hyperline segment defined by two end points with a corresponding membership function. The modification of intersection test in the MFHLSCNN has resulted in improved performance. The thyroid dataset is taken from UCI machine learning repository. The results obtained with the proposed approach are compared with the multilayer perceptron (MLP trained using error backpropagation and FHLSCNN. The experimental results show that performance of the proposed approach is superior as compared to MLP and FHLSCNN. Moreover training time and recall time per pattern of MFHLSCNN and FHLSCNN is very less as compared to MLP.

Satish N. Kulkarni

2012-12-01

348

Evaluation of oil thickness by neural network analysis of IR imagery

International Nuclear Information System (INIS)

The feasibility of using neural network analysis of conventional thermal infra-red data gathered from surveillance aircraft to determine the thickness of oil at sea, was examined. Sea trial data was examined using Multi-Layer Perceptron neural network architecture, based on indications that it was the most appropriate configuration for determining oil thickness. Core input variables included oil brightness, time of day, sea brightness, wind speed, oil type, and sea temperature. Other variables, such as altitude, wave height, air temperature, camera gain, and others, did not appear to produce any significant difference in the prediction performance. By using only a restricted sea trial data set in training the network, it was found that it was possible to correctly classify about 80 per cent of the data into one of four thickness classes. Since there was no additional data available to validate the network, these results were considered encouraging, but not definitive. Additional data will be collected in planned future sea trials to further evaluate the accuracy of the trained network. 4 refs., 6 tabs., 4 figs

349

The preliminary study presented within this paper shows a comparative study of various texture features extracted from liver ultrasonic images by employing Multilayer Perceptron (MLP), a type of artificial neural network, to study the presence of disease conditions. An ultrasound (US) image shows echo-texture patterns, which defines the organ characteristics. Ultrasound images of liver disease conditions such as “fatty liver,” “cirrhosis,” and “hepatomegaly” produce distinctive echo patterns. However, various ultrasound imaging artifacts and speckle noise make these echo-texture patterns difficult to identify and often hard to distinguish visually. Here, based on the extracted features from the ultrasonic images, we employed an artificial neural network for the diagnosis of disease conditions in liver and finding of the best classifier that distinguishes between abnormal and normal conditions of the liver. Comparison of the overall performance of all the feature classifiers concluded that “mixed feature set” is the best feature set. It showed an excellent rate of accuracy for the training data set. The gray level run length matrix (GLRLM) feature shows better results when the network was tested against unknown data.

Lele, Ramachandra Dattatraya; Joshi, Mukund; Chowdhary, Abhay

2014-01-01

350

International Nuclear Information System (INIS)

An artificial neural network model was developed for modeling of the effects of mechanical alloying parameters including milling time, milling speed and ball to powder weight ratio on the characteristics of Al-8 vol%SiC nanocomposite powders. The crystallite size and lattice strain of the aluminum matrix were considered for modeling. This nanostructured nanocomposite powder was synthesized by utilizing planetary high energy ball mill and the required data for training were collected from the experimental results. The characteristics of the particles were determined by X-ray diffraction, scanning and transmission electron microscopy. Two types of neural network architecture, i.e. multi-layer perceptron (MLP) and radial basis function (RBF), were used. The steepest descent along with variable learning rate back-propagation algorithm known as a heuristic technique was utilized for training the MLP network. It was found that MLP network yields better results compared to RBF network, giving an acceptable mapping between the network responses and the target data with a high correlation coefficients. The response surfaces between the response variables, i.e. crystallite size, lattice strain of the aluminum matrix and the processing parameters are presented. The procedure modeling can be used for optimization of the MA process for synthesizing of nanostructured metal matrix nanocomposites

351

International Nuclear Information System (INIS)

Numerical simulation of natural circulation boiling water reactor is important in order to study its performance for different designs and under various off-design conditions. Numerical simulations can be performed by using thermal-hydraulic codes. Very fast numerical simulations, useful for extensive parametric studies and for solving design optimization problems, can be achieved by using an artificial neural network (ANN) model of the system. In the present work, numerical simulations of natural circulation boiling water reactor have been performed with RELAP5 code for different values of design parameters and operational conditions. Parametric trends observed have been discussed. The data obtained from these simulations have been used to train artificial neural networks, which in turn have been used for further parametric studies and design optimization. The ANN models showed error within ±5% for all the simulated data. Two most popular methods, multilayer perceptron (MLP) and radial basis function (RBF) networks, have been used for the training of ANN model. Sequential quadratic programming (SQP) has been used for optimization

352

International Nuclear Information System (INIS)

A nuclear power plant's (NPP's) status is usually monitored by a human operator. Any classifier system used to enhance the operator's capability to diagnose a safety-critical system like an NPP should classify a novel transient as ''don't-know'' if it is not contained within its accumulated knowledge base. In particular, the classifier needs some kind of proximity measure between the new data and its training set. Artificial neural networks have been proposed as NPP classifiers, the most popular ones being the multilayered perceptron (MLP) type. However, MLPs do not have a proximity measure, while learning vector quantization, probabilistic neural networks (PNNs), and some others do. This proximity measure may also serve as an explanation to the classifier's decision in the way that case-based-reasoning expert systems do. The capability of a PNN network as a classifier is demonstrated using simulator data for the three-loop 436-MW(electric) Westinghouse San Onofre unit 1 pressurized water reactor. A transient's classification history is used in an ''evidence accumulation'' technique to enhance a classifier's accuracy as well as its consistency

353

On-line control of the COMPASS-D tokamak using a neural network

International Nuclear Information System (INIS)

Multi-layer perceptron (MLP) networks are particularly appropriate for performing rapid non-linear mapping. In the application discussed in this Paper the position and shape of the plasma within the experimental fusion research tokamak COMPASS-D at UKAEA's Culham Laboratory is determined from a series of magnetic sensors placed around the vacuum vessel, close to the plasma boundary. By using a real-time analogue neural network it is possible to achieve control within a sub-millisecond time-scale. In this application the neural network is needed to solve an inverse problem. Numerical codes exist that are able to calculate the signals expected on the magnetic sensors for a given plasma position and profile. The problem is well defined from the solution of the Grad-Shafranov equation. However, no easy analytical formalism exists to reverse the problem - to calculate the plasma parameters given the magnetic signals. It is this mapping, from the set of magnetic diagnostic input signals to the parameters of the plasma, that an MLP network can be trained to solve. The training data are some 2000 example plasma equilibria, covering the likely possible configurations of the plasma, solved by numerical methods. The desired aim, to control the plasma boundary position to within a few millimetres, has now been achieved. (author)

354

The computer-aided diagnosis system we developed for the mass characterization is mainly based on a segmentation algorithm and on the neural classification of several features computed on the segmented mass. Mass segmentation plays a key role in most computerized systems. Our technique is a gradient-based one, showing the main characteristic that no free parameters have been evaluated on the dataset used in this analysis, thus it can directly be applied to datasets acquired in different conditions without any ad-hoc modification. A dataset of 226 masses (109 malignant and 117 benign) has been used in this study. The segmentation algorithm works with a comparable efficiency both on malignant and benign masses. Sixteen features based on shape, size and intensity of the segmented masses are analyzed by a multi-layered perceptron neural network. A feature selection procedure has been carried out on the basis of the feature discriminating power and of the linear correlations interplaying among them. The comparison...

Delogu, P; Kasae, P; Retico, A

2008-01-01

355

The quaternionic singular value decomposition is a technique to decompose a quaternion matrix (representation of a colour image) into quaternion singular vector and singular value component matrices exposing useful properties. The objective of this study was to use a small portion of uncorrelated singular values, as robust features for the classification of sliced pork ham images, using a supervised artificial neural network classifier. Images were acquired from four qualities of sliced cooked pork ham typically consumed in Ireland (90 slices per quality), having similar appearances. Mahalanobis distances and Pearson product moment correlations were used for feature selection. Six highly discriminating features were used as input to train the neural network. An adaptive feedforward multilayer perceptron classifier was employed to obtain a suitable mapping from the input dataset. The overall correct classification performance for the training, validation and test set were 90.3%, 94.4%, and 86.1%, respectively. The results confirm that the classification performance was satisfactory. Extracting the most informative features led to the recognition of a set of different but visually quite similar textural patterns based on quaternionic singular values. PMID:20374805

Valous, Nektarios A; Mendoza, Fernando; Sun, Da-Wen; Allen, Paul

2010-03-01

356

Neural and fuzzy computation techniques for playout delay adaptation in VoIP networks.

Playout delay adaptation algorithms are often used in real time voice communication over packet-switched networks to counteract the effects of network jitter at the receiver. Whilst the conventional algorithms developed for silence-suppressed speech transmission focused on preserving the relative temporal structure of speech frames/packets within a talkspurt (intertalkspurt adaptation), more recently developed algorithms strive to achieve better quality by allowing for playout delay adaptation within a talkspurt (intratalkspurt adaptation). The adaptation algorithms, both intertalkspurt and intratalkspurt based, rely on short term estimations of the characteristics of network delay that would be experienced by up-coming voice packets. The use of novel neural networks and fuzzy systems as estimators of network delay characteristics are presented in this paper. Their performance is analyzed in comparison with a number of traditional techniques for both inter and intratalkspurt adaptation paradigms. The design of a novel fuzzy trend analyzer system (FTAS) for network delay trend analysis and its usage in intratalkspurt playout delay adaptation are presented in greater detail. The performance of the proposed mechanism is analyzed based on measured Internet delays. Index Terms-Fuzzy delay trend analysis, intertalkspurt, intratalkspurt, multilayer perceptrons (MLPs), network delay estimation, playout buffering, playout delay adaptation, time delay neural networks (TDNNs), voice over Internet protocol (VoIP). PMID:16252825

Ranganathan, Mohan Krishna; Kilmartin, Liam

2005-09-01

357

A telemedicine system using communication and information technology to deliver medical signals such as ECG, EEG for long distance medical services has become reality. In either the urgent treatment or ordinary healthcare, it is necessary to compress these signals for the efficient use of bandwidth. This paper discusses a quality on demand compression of EEG signals using neural network predictors for telemedicine applications. The objective is to obtain a greater compression gains at a low bit rate while preserving the clinical information content. A two-stage compression scheme with a predictor and an entropy encoder is used. The residue signals obtained after prediction is first thresholded using various levels of thresholds and are further quantized and then encoded using an arithmetic encoder. Three neural network models, single-layer and multi-layer perceptrons and Elman network are used and the results are compared with linear predictors such as FIR filters and AR modeling. The fidelity of the reconstructed EEG signal is assessed quantitatively using parameters such as PRD, SNR, cross correlation and power spectral density. It is found from the results that the quality of the reconstructed signal is preserved at a low PRD thereby yielding better compression results compared to results obtained using lossless scheme. PMID:21785587

Sriraam, N

2011-01-01

358

Estimation of mean grain size of seafloor sediments using neural network

The feasibility of an artificial neural network based approach is investigated to estimate the values of mean grain size of seafloor sediments using four dominant echo features, extracted from acoustic backscatter data. The acoustic backscatter data were collected using a dual-frequency (33 and 210 kHz) single-beam, normal-incidence echo sounder at twenty locations in the central part of the western continental shelf of India. Statistically significant correlations are observed between the estimated average values of mean grain size of sediments and the ground-truth data at both the frequencies. The results indicate that once a multi-layer perceptron model is trained with back-propagation algorithm, the values of mean grain size can reasonably be estimated in an experimental area. The study also revealed that the consistency among the estimated values of mean grain size at different acoustic frequencies is considerably improved with the neural network based method as compared to that with a model-based approach.

de, Chanchal; Chakraborty, Bishwajit

2012-03-01

359

Object segmentation is the process by which a mask is generated which identifies the area of an image which is occupied by an object. Many object recognition techniques depend on the quality of such masks for shape and underlying brightness information, however, segmentation remains notoriously unreliable. This paper considers how the image restoration technique of Geman and Geman can be applied to the improvement of object segmentations generated by a locally adaptive background subtraction technique. Also presented is how an artificial neural network hybrid, consisting of a single layer Kohonen network with each of its nodes connected to a different multi-layer perceptron, can be used to approximate the image restoration process. It is shown that the restoration techniques are very well suited for parallel processing and in particular the artificial neural network hybrid has the potential for near real time image processing. Results are presented for the detection of ships in SPOT panchromatic imagery and the detection of vehicles in infrared linescan images, these being a fair representation of the wider class of problem.

Foulkes, Stephen B.; Booth, David M.

1997-07-01

360

Bayesian regularization of artificial neural networks (BRANNs) were used to predict body mass index (BMI) in mice using single nucleotide polymorphism (SNP) markers. Data from 1896 animals with both phenotypic and genotypic (12 320 loci) information were used for the analysis. Missing genotypes were imputed based on estimated allelic frequencies, with no attempt to reconstruct haplotypes based on family information or linkage disequilibrium between markers. A feed-forward multilayer perceptron network consisting of a single output layer and one hidden layer was used. Training of the neural network was done using the Bayesian regularized backpropagation algorithm. When the number of neurons in the hidden layer was increased, the number of effective parameters, ?, increased up to a point and stabilized thereafter. A model with five neurons in the hidden layer produced a value of ? that saturated the data. In terms of predictive ability, a network with five neurons in the hidden layer attained the smallest error and highest correlation in the test data although differences among networks were negligible. Using inherent weight information of BRANN with different number of neurons in the hidden layer, it was observed that 17 SNPs had a larger impact on the network, indicating their possible relevance in prediction of BMI. It is concluded that BRANN may be at least as useful as other methods for high-dimensional genome-enabled prediction, with the advantage of its potential ability of capturing non-linear relationships, which may be useful in the study of quantitative traits under complex gene action. PMID:21481292

Okut, Hayrettin; Gianola, Daniel; Rosa, Guilherme J M; Weigel, Kent A

2011-06-01

361

Foreground removal from CMB temperature maps using an MLP neural network

One of the main obstacles for extracting the Cosmic Microwave Background (CMB) signal from observations in the mm-submm range is the foreground contamination by emission from Galactic components: mainly synchrotron, free-free and thermal dust emission. Due to the statistical nature of the intrinsic CMB signal it is essential to minimize the systematic errors in the CMB temperature determinations. Following the available knowledge of the spectral behavior of the Galactic foregrounds simple power law-like spectra have been assumed. The feasibility of using a simple neural network for extracting the CMB temperature signal from the combined signal CMB and the foregrounds has been investigated. As a specific example, we have analysed simulated data, as expected from the ESA Planck CMB mission. A simple multilayer perceptron neural network with 2 hidden layers can provide temperature estimates over more than 80 per cent of the sky that are to a high degree uncorrelated with the foreground signals. A single network will be able to cover the dynamic range of the Planck noise level over the entire sky.

Nørgaard-Nielsen, H. U.; Jørgensen, H. E.

2008-12-01

362

Investigation of efficient features for image recognition by neural networks.

In the paper, effective and simple features for image recognition (named LiRA-features) are investigated in the task of handwritten digit recognition. Two neural network classifiers are considered-a modified 3-layer perceptron LiRA and a modular assembly neural network. A method of feature selection is proposed that analyses connection weights formed in the preliminary learning process of a neural network classifier. In the experiments using the MNIST database of handwritten digits, the feature selection procedure allows reduction of feature number (from 60 000 to 7000) preserving comparable recognition capability while accelerating computations. Experimental comparison between the LiRA perceptron and the modular assembly neural network is accomplished, which shows that recognition capability of the modular assembly neural network is somewhat better. PMID:22391231

Goltsev, Alexander; Gritsenko, Vladimir

2012-04-01

363

Directory of Open Access Journals (Sweden)

Full Text Available En este artículo, se modela el precio promedio mensual del café colombiano en la Bolsa de Nueva York, usando varios modelos alternativos. El modelo final seleccionado está compuesto por una componente lineal autorregresiva más una red neuronal artificial tipo perceptron multicapa con dos neuronas en la capa oculta, que permite representar la dinámica que sigue el valor esperado de la serie de precios; mientras que la dinámica de los residuales es especificada usando un proceso heterocedástico condicional autoregresivo de primer orden. Los residuales normalizados del modelo son incorrelacionados y homocedásticos, y siguen aproximadamente una distribución normal. Los resultados indican que el precio actual depende de los precios ocurridos en los últimos cuatro meses.In this paper, the monthly average price of the Colombian coffee in the New York Stock Exchange, is modelling by means of several alternative models. The preferred model is composed by a lineal autoregressive component plus a multilayer perceptron neural network with two neurons in the hidden layer, that allow us to representing the dynamic following by the expected value of the price time series; while, the dynamic of the residuals is specified by an autoregressive conditional heterocedastic model of first order. The normalized residuals of the preferred model are uncorrelated, homocedastic and are distributed following a normal distribution. The results indicate that the current price depend of the prices in the previous four months.

Juan David Velásquez Henao

2007-12-01

364

Scientific Electronic Library Online (English)

Full Text Available SciELO Colombia | Language: Spanish Abstract in spanish Un nuevo modelo híbrido es propuesto para pronosticar el índice colombiano de precios al consumidor. Este es basado en una descomposición estructural de la serie temporal con el objetivo de remover cualquier patrón fácilmente detectable en los datos, y en el uso de un perceptron multicapa para model [...] ar las relaciones ocultas en la serie de tiempo. Los resultados superan las aproximaciones clásicas basadas en la aproximación de Box y Jenkins, y los modelos convencionales de Redes Neuronales, e incentivan el estudio de este tipo de aproximación híbrida para modelar otras series temporales. Abstract in english A new hybrid model is proposed to forecasting the Colombian Consumer Price Index. It’s based on the structural decomposition of the original time series with the aim of remove any easily detected pattern in the data, and in the use of multilayer perceptron to model hidden relationships in the studie [...] d time series. The results overcome classical approaches based on Box-Jenkins methodology and conventional neural networks methodology, and encourage the study of this hybrid approach to modelling other time series.

JUAN DAVID, VELÁSQUEZ HENAO; SANTIAGO FERNANDO, MONTOYA MORENO.

365

Scientific Electronic Library Online (English)

Full Text Available SciELO Colombia | Language: Spanish Abstract in spanish Un nuevo modelo híbrido es propuesto para pronosticar el índice colombiano de precios al consumidor. Este es basado en una descomposición estructural de la serie temporal con el objetivo de remover cualquier patrón fácilmente detectable en los datos, y en el uso de un perceptron multicapa para model [...] ar las relaciones ocultas en la serie de tiempo. Los resultados superan las aproximaciones clásicas basadas en la aproximación de Box y Jenkins, y los modelos convencionales de Redes Neuronales, e incentivan el estudio de este tipo de aproximación híbrida para modelar otras series temporales. Abstract in english A new hybrid model is proposed to forecasting the Colombian Consumer Price Index. It’s based on the structural decomposition of the original time series with the aim of remove any easily detected pattern in the data, and in the use of multilayer perceptron to model hidden relationships in the studie [...] d time series. The results overcome classical approaches based on Box-Jenkins methodology and conventional neural networks methodology, and encourage the study of this hybrid approach to modelling other time series.

JUAN DAVID, VELÁSQUEZ HENAO; SANTIAGO FERNANDO, MONTOYA MORENO.

2005-11-01

366

Neural network method applied to particle image velocimetry

The last two decades have seen rapid developments in computing taking as their inspiration the human brain. The human brain functions in a highly parallel and distributed fashion. The adaptive structure of the brain means that learning or training can accompany decision making. This basic neural model has inspired computer hardware exhibiting a parallelism which has revolutionised processing speeds in complex task analysis. Similarly there has been substantial activity in the field of intelligent software and in particular in the area ofneural computing. The human brain may viewed as composed of approximately 1 dbasic units, the neurons. Each neuron exhibits a high degree of interconnectivity with connections to approximately 1 O other neurons. Each neuron accepts many inputs which are added or integrated in some fashion and this causes the neuron to become active or passive. The active neuron emits an output to interconnected neurons. The importance of any one input is controlled by the effectiveness of the corresponding interconnection or weight. One area that has attracted attention in the application of neural networks is pattern recognition. Here the functions of feature classification and extraction are handled by a network which receives some education or training prior to the task of recognition. A priori knowledge of expected outcomes is used as a starting point with the network being allowed to modify or enlarge its knowledge base as the task proceeds. Various models or approaches to adaptive problem solving have been developed. The pattern recognition problem considered in the present paper is the identification of image grouping in double exposure PIV images. The aim is to provide an adaptive net which, following initial training, is able to identify image partners and adapt to changing flow conditions. This latter feature is seen as essential in order that the full potential of the neural net in temporally or spatially changing flow regimes can be realised. An important class of neural network is the multi-layer perceptron. The neurons are distributed on surfaces and linked by weighted interconnections. In the present paper we demonstrate how this type of net can developed into a competitive, adaptive filter which will identify PIV image pairs in a number of commonly occurring flow types. Previous work by the authors in particle tracking analysis (1, 2) has shown the efficiency of statistical windowing techniques in flows without systematic (in time or space) variations. The effectiveness of the present neural net is illustrated by applying it to digital simulations ofturbulent and rotating flows. Work reported by Cenedese et al (3) has taken a different approach in examining the potential for neural net methods applied to PIV.

Grant, Ian; Pan, X.

1993-12-01

367

Handwritten Arabic Numeral Recognition using a Multi Layer Perceptron

Handwritten numeral recognition is in general a benchmark problem of Pattern Recognition and Artificial Intelligence. Compared to the problem of printed numeral recognition, the problem of handwritten numeral recognition is compounded due to variations in shapes and sizes of handwritten characters. Considering all these, the problem of handwritten numeral recognition is addressed under the present work in respect to handwritten Arabic numerals. Arabic is spoken throughout the Arab World and the fifth most popular language in the world slightly before Portuguese and Bengali. For the present work, we have developed a feature set of 88 features is designed to represent samples of handwritten Arabic numerals for this work. It includes 72 shadow and 16 octant features. A Multi Layer Perceptron (MLP) based classifier is used here for recognition handwritten Arabic digits represented with the said feature set. On experimentation with a database of 3000 samples, the technique yields an average recognition rate of 94....

Das, Nibaran; Saha, Sudip; Haque, Syed Sahidul

2010-01-01

368

This paper investigates the potential application of neural network to inversion of soil moisture using polarimetric remote sensing data. The neural network used for the inversion of soil parameters is multi-layer perceptron trained with the back-propagation algorithm. The training data include the polarimetric backscattering coefficients obtained from theoretical surface scattering models together with an assumed nominal range of soil parameters which are comprised of the soil permittivity and surface roughness parameters. Soil permittivity is calculated from the soil moisture and the assumed soil texture based on an empirical formula at C-, L-, and P-bands. The rough surface parameters for the soil surface, which is described by the Gaussian random process, are the root-mean-square (rms) height and correlation length. For the rough surface scattering, small perturbation method is used for the L-band frequency, and Kirchhoff approximation is used for the C-band frequency to obtain the corresponding backscattering coefficients. During the training, the backscattering coefficients are the inputs to the neural net and the output from the net are compared with the desired soil parameters to adjust the interconnecting weights. The process is repeated for each input-output data entry and then for the entire training data until convergence is reached. After training, the backscattering coefficients are applied to the trained neural net to retrieve the soil parameters which are compared with the desired soil parameters to verify the effectiveness of this technique. Several cases are examined. First, for simplicity, the correlation length and rms height of the soil surface are fixed while soil moisture is varied. Soil moisture obtained using the neural networks with either L-band or C-band backscattering coefficients for the HH and VV polarizations as inputs is in good agreement with the desired soil moisture. The neural net output matches the desired output for the soil moisture range of 16 to 60 percent for the C-band case. The next case investigated is to vary both soil moisture and rms height while keeping the correlation length fixed. For this case, C-band backscattering coefficients are not sufficient for retrieving two parameters because the Kirchhoff approximation gives the same HH and VV backscattering coefficients. Therefore, the backscattering coefficients at two different frequency bands are necessary to find both the soil moisture and rms height. Finally, the neural nets are also applied to simultaneously invert soil moisture, rms height, and correlation length. Overall, the soil moisture retrieved from the neural network agrees very well with the desired soil moisture. This suggests that the neural network shows potential for retrieval of soil parameters from remote sensing data.

Wang, L.; Shin, R. T.; Kong, J. A.; Yueh, S. H.

1993-01-01

369

Neural network sensor fusion: Creation of a virtual sensor for cloud-base height estimation

Sensor fusion has become a significant area of signal processing research that draws on a variety of tools. Its goals are many, however in this thesis, the creation of a virtual sensor is paramount. In particular, neural networks are used to simulate the output of a LIDAR (LASER. RADAR) that measures cloud-base height. Eye-safe LIDAR is more accurate than the standard tool that would be used for such measurement; the ceilometer. The desire is to make cloud-base height information available at a network of ground-based meteorological stations without actually installing LIDAR sensors. To accomplish this, fifty-seven sensors ranging from multispectral satellite information to standard atmospheric measurements such as temperature and humidity, are fused in what can only be termed as a very complex, nonlinear environment. The result is an accurate prediction of cloud-base height. Thus, a virtual sensor is created. A total of four different learning algorithms were studied; two global and two local. In each case, the very best state-of-the-art learning algorithms have been selected. Local methods investigated are the regularized radial basis function network, and the support vector machine. Global methods include the standard backpropagation with momentum trained multilayer perceptron (used as a benchmark) and the multilayer perceptron trained via the Kalman filter algorithm. While accuracy is the primary concern, computational considerations potentially limit the application of several of the above techniques. Thus, in all cases care was taken to minimize computational cost. For example in the case of the support vector machine, a method of partitioning the problem in order to reduce memory requirements and make the optimization over a large data set feasible was employed and in the Kalman algorithm case, node-decoupling was used to dramatically reduce the number of operations required. Overall, the methods produced somewhat equivalent mean squared errors indicating that the descriptive capacity of the data had been reached. However, the support vector machine was the clear winner in terms of computational complexity. As well, through its ability to determine its own dimensionality it is able to relate information about the physics of the problem back to the user. This thesis, contributes to the literature on three fronts. First, it demonstrates the concept of creating of a virtual sensor via sensor fusion. Second, in the remote-sensing field where focus has typically been on pattern classification tasks, this thesis provides an in-depth look at the use of neural networks for tough regression problems. And lastly, it provides a useful tool for the meteorological community in creating the ability to add large-scale, cloud-field information to predictive models.

Pasika, Hugh Joseph Christopher

2000-10-01

370

Regional TEC mapping using neural networks

Characterization and modeling of ionospheric variability in space and time is very important for communications and navigation. To characterize the variations, the ionosphere should be monitored, and the sparsity of the measurements has to be compensated by interpolation algorithms. The total electron content (TEC) is a major parameter that can be used to obtain regional ionospheric maps. In this study, neural networks (NNs), specifically multilayer perceptrons (MLPs) and radial basis function networks (RBFN), are investigated for the merits of their nonlinear modeling capability. In order to assess the performance of MLP and RBFN structures with respect to mapping and ionospheric parameters, these algorithms are applied to synthetically generated TEC surfaces representing various ionospheric states. Synthetic TEC data are sampled homogenously and randomly for a varying number of data points. The reconstruction errors show that the performance improves significantly when homogenous sampling is preferred to random station distribution. The best MLP and RBFN structures for any possible realistic scenario are determined by examining the performance parameters for all possible cases. It is also observed that RBFN with local receptive fields relies more on the number of training data points. In contrast to RBFN, MLP as a global approximator depends strongly on ionospheric trends. Finally, chosen MLP and RBFN models are applied to a set of real GPS-TEC values obtained from central Europe, and their performances are compared with well known Global Ionospheric Maps produced by the International GNSS Service. The resolution and interpolation quality of the generated maps indicate that NNs offer a powerful and reliable alternative to the conventional TEC mapping algorithms.

Yilmaz, A.; Akdogan, K. E.; Gurun, M.

2009-06-01

371

Directory of Open Access Journals (Sweden)

Full Text Available The increased use of nonlinear devices in the industry has resulted in the direct increase of harmonic distortion in power systems during these last years. Active filter systems are proposed to mitigate current harmonics generated by nonlinear loads. The conventional scheme based on a two-level voltage source inverter controlled by a hysteresis controller has several disadvantages and cannot be used for medium or high-power applications. To overcome these drawbacks and improve the APF performance, there’s a great tendency to use multilevel inverters controlled by intelligent controllers. Three level (NPC inverter is one of the most widely used topologies in various industrial applications such as machine drives and power factor compensators. On the other hand, artificial neural networks are under study and investigation in other power electronics applications. In order to gain the advantages of the three-level inverter and artificial neural networks and to reduce the complexity of classical control schemes, a new active power filter configuration controlled by two MLPNN (Multi-Layer Perceptron Neural Network is proposed in this paper. The first ANN is used to replace the PWM current controller, and the second one to maintain a constant dc link voltage across the capacitors and compensate the inverter power losses. The performance of the global system, including power and control circuits is evaluated by Matlab-Simulink and SimPowerSystem Toolbox simulation. The obtained results confirm the effectiveness of the proposed control scheme.

Chennai Salim

2011-09-01

372

Modeling of gamma-ray energy absorption buildup factors using neural network

International Nuclear Information System (INIS)

This paper presents a new approach based on multilayered perceptrons (MLPs) to compute energy absorption buildup factors. The MLP has been trained by a Levenberg-Marquardt learning algorithm. The model is fast and does not require tremendous computational efforts. The results obtained by using the proposed model are in good agreement with the ANSI/ANS-6.4.3 standard data set

373

Scientific Electronic Library Online (English)

Full Text Available SciELO Colombia | Language: Spanish Abstract in spanish En este trabajo se presenta una red neuronal perceptron multicapa combinada con el método Nelder-Mead Simplex para detectar daño en vigas. Los parámetros de entrada a la red se basan en frecuencias naturales y flexibilidad modal. Se considera que solo una cantidad específica de modos fueron identifi [...] cados y que se dispone de mediciones en grados de libertad verticales. La confiabilidad de la metodología propuesta se evalúa a partir de escenarios de daño aleatorios y de la definición de 3 tipos de error que la red puede cometer durante el proceso de detección del daño. Los resultados muestran que la metodología puede determinar confiablemente los escenarios de daño buscados. Sin embargo, su aplicación a vigas de gran tamaño puede verse limitada por el elevado costo computacional asociado al entrenamiento de la red. Abstract in english In this paper is presented a multilayer perceptron neural network combined with the Nelder-Mead Simplex method to detect damage in multiple support beams. The input parameters are based on natural frequencies and modal flexibility. It was considered that only a number of modes were available and tha [...] t only vertical degrees of freedom were measured. The reliability of the proposed methodology is assessed from the generation of random damages scenarios and the definition of three types of errors, which can be found during the damage identification process. Results show that the methodology can reliably determine the damage scenarios. However, its application to large beams may be limited by the high computational cost of training the neural network.

Jesús D., Villalba; Ivan D., Gómez; José E., Laier.

2012-06-01

374

Nighttime cloud properties retrieval using MODIS and artificial neural networks

The aim of this work is to develop a methodology for inferring water cloud macro and microphysical properties from nighttime MODIS imagery This method is based on the inversion of a theoretical radiative transfer model that simulates the radiances detected in each of the sensor infrared bands In this case LibRadtran package Mayer and Kylling 2005 was used which allows us the calculation of the radiation field in the Earth s atmosphere given a specified set of atmospheric and cloud parameters However due to the complexity of this forward model its inversion cannot be performed in an analytical way To accomplish this task we propose an operational technique based on artificial neural networks ANNs whose main characteristic is the ability to retrieve cloud properties much faster than conventional methods Platnick et al 2003 Gonzalez et al 2002 Thus the procedure is as follows Using the theoretical radiative model a Look Up Table LUT is generated for a great variety of surface cloud and atmospheric conditions This dataset is divided randomly into a training set two-thirds of the items and a test set one third of the items which are used to train the neural network in order to fit the inversion problem In this study multilayer perceptrons MLPs with two hidden layers are used and the backpropagation with momentum method is used in the training process Furthermore to accelerate the convergence of ANN s evolutionary techniques are used to search the ANN configuration that provides the best fit Furthermore in order to check the

Pérez, J. C.; Cerdeña, A.; González, A.

375

Committee neural network model for rock permeability prediction

Quantitative formulation between conventional well log data and rock permeability, undoubtedly the most critical parameter of hydrocarbon reservoir, could be a potent tool for solving problems associated with almost all tasks involved in petroleum engineering. The present study proposes a novel approach in charge of the quest for high-accuracy method of permeability prediction. At the first stage, overlapping of conventional well log data (inputs) was eliminated by means of principal component analysis (PCA). Subsequently, rock permeability was predicted from extracted PCs using multi-layer perceptron (MLP), radial basis function (RBF), and generalized regression neural network (GRNN). Eventually, a committee neural network (CNN) was constructed by virtue of genetic algorithm (GA) to enhance the precision of ultimate permeability prediction. The values of rock permeability, derived from the MPL, RBF, and GRNN models, were used as inputs of CNN. The proposed CNN combines results of different ANNs to reap beneficial advantages of all models and consequently producing more accurate estimations. The GA, embedded in the structure of the CNN assigns a weight factor to each ANN which shows relative involvement of each ANN in overall prediction of rock permeability from PCs of conventional well logs. The proposed methodology was applied in Kangan and Dalan Formations, which are the major carbonate reservoir rocks of South Pars Gas Field-Iran. A group of 350 data points was used to establish the CNN model, and a group of 245 data points was employed to assess the reliability of constructed CNN model. Results showed that the CNN method performed better than individual intelligent systems performing alone.

Bagheripour, Parisa

2014-05-01

376

In the context of detection of weeds in crops for site-specific weed control, on-ground spectral reflectance measurements are the first step to determine the potential of remote spectral data to classify weeds and crops. Field studies were conducted for four years at different locations in Spain. We aimed to distinguish cruciferous weeds in wheat and broad bean crops, using hyperspectral and multispectral readings in the visible and near-infrared spectrum. To identify differences in reflectance between cruciferous weeds, we applied three classification methods: stepwise discriminant (STEPDISC) analysis and two neural networks, specifically, multilayer perceptron (MLP) and radial basis function (RBF). Hyperspectral and multispectral signatures of cruciferous weeds, and wheat and broad bean crops can be classified using STEPDISC analysis, and MLP and RBF neural networks with different success, being the MLP model the most accurate with 100%, or higher than 98.1%, of classification performance for all the years. Classification accuracy from hyperspectral signatures was similar to that from multispectral and spectral indices, suggesting that little advantage would be obtained by using more expensive airborne hyperspectral imagery. Therefore, for next investigations, we recommend using multispectral remote imagery to explore whether they can potentially discriminate these weeds and crops. PMID:22629171

de Castro, Ana-Isabel; Jurado-Expósito, Montserrat; Gómez-Casero, María-Teresa; López-Granados, Francisca

2012-01-01

377

The infamous soils of Adapazari, Turkey, that failed extensively during the 46-s long magnitude 7.4 earthquake in 1999 have since been the subject of a research program. Boreholes, piezocone soundings and voluminous laboratory testing have enabled researchers to apply sophisticated methods to determine the soil profiles in the city using the existing database. This paper describes the use of the artificial neural network (ANN) model to predict the complex soil profiles of Adapazari, based on cone penetration test (CPT) results. More than 3236 field CPT readings have been collected from 117 soundings spread over an area of 26 km2. An attempt has been made to develop the ANN model using multilayer perceptrons trained with a feed-forward back-propagation algorithm. The results show that the ANN model is fairly accurate in predicting complex soil profiles. Soil identification using CPT test results has principally been based on the Robertson charts. Applying neural network systems using the chart offers a powerful and rapid route to reliable prediction of the soil profiles.

Arel, Ersin

2012-06-01

378

A neural network structure has been used for unfolding neutron spectra measured by means of a Bonner Sphere Spectrometer set and a foil activation set using several neutron induced reactions. The present work used the SNNS (Stuttgart Neural Network Simulator) as the interface for designing, training and validation of the Multilayer Perceptron network. The back-propagation algorithm was applied. The Bonner Sphere set chosen has been calibrated at the National Physical Laboratory, United Kingdom, and uses gold activation foils as thermal neutron detectors. The neutron energy covered by the response functions goes from 0.0001 eV to 14 MeV. The foil activation set chosen has been irradiated at the IEA-R1 research reactor and measured at the Nuclear Metrology Laboratory of IPEN-CNEN/SP. Two types of neutron spectra were numerically investigated: monoenergetic and continuous The unfolded spectra were compared to a conventional method using code SAND-II as part of the neutron dosimetry system SAIPS. Good results wer...

Braga, C C

2001-01-01

379

using artificial neural network

Directory of Open Access Journals (Sweden)

Full Text Available In this work, a Multilayer Perceptron implementation ? MLP using functional Magnetic Resonance Imaging (fMRI is used to infer stimuli performed. Sets of images of brain activation were generated by visual, auditory and finger tapping paradigms in 54 healthy volunteers. These images were used for training the MLP network in a leave-one-out manner in order to predict the paradigm that a subject performed by using other images, so far unseen by the MLP network. The aim in this paper is the exploring of the influence of the number of the Principal Component (PC on the performance of the MLP in classifying fMRI paradigms. The classifier´s performance was evaluated in terms of the Sensitivity and Specificity, Prediction Accuracy and the area Az under the receiver operating characteristics (ROC curve. From the ROC analysis, values of Az up to 1 were obtained with 60 PCs in discriminating the visual paradigm from the auditory paradigm.

Rafael do Esp\\u00EDrito Santo

2007-01-01

380

Modular Network SOM (mnSOM: A New Powerful Tool in Neural Networks

Directory of Open Access Journals (Sweden)

Full Text Available In this paper, a new powerful method in artificial neural networks, called modular network SOM (mnSOM is introduced. mnSOM is a generalization of Self Organizing Maps (SOM formed by replacing each vector unit of SOM with function module. The modular function could be a multi layer perceptron, a recurrent neural network or even SOM itself. Having this flexibility, mnSOM becomes a new powerful tool in artificial neural network.

Muhammad Aziz Muslim

2009-12-01

381

Modular Network SOM (mnSOM): A New Powerful Tool in Neural Networks

Digital Repository Infrastructure Vision for European Research (DRIVER)

In this paper, a new powerful method in artificial neural networks, called modular network SOM (mnSOM) is introduced. mnSOM is a generalization of Self Organizing Maps (SOM) formed by replacing each vector unit of SOM with function module. The modular function could be a multi layer perceptron, a recurrent neural network or even SOM itself. Having this flexibility, mnSOM becomes a new powerful tool in artificial neural network.

Muhammad Aziz Muslim

2009-01-01

382

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: Portuguese Abstract in portuguese Estudar modelagens através de dados geodésicos temporais com a possibilidade de predizer a posição de linha de costa é uma tarefa importante e pode auxiliar significativamente na gestão costeira. A área de estudo neste trabalho se refere ao município de Matinhos no estado do Paraná, Brasil. As linha [...] s de costa temporais utilizadas para testar a modelagem preditiva são provenientes respectivamente da fotogrametria analógica para anos 1954, 1963, 1980, 1991 e 1997 e de levantamentos geodésicos utilizando GPS (Global Position System) para 2001, 2002, 2005 e 2008 (como controle). Dois testes com as redes neurais artificiais foram organizados mudando alguns parâmetros como: arquitetura, número de neurônios nas camadas ocultas e algoritmos de treinamentos. Quando comparados o valor dos resíduos entre a predição e a linha de costa de controle, os melhores resultados estatísticos indicam que o MAPE (mean absolute percentage error) são 0,28% utilizando a rede neural parcialmente recorrente de Elman com o algoritmo de treinamento quase-Newton e 0,46% para o caso da rede neural perceptron multicamadas com o algoritmo de treinamento utilizando o método Bayesiano com regularização. Abstract in english The study of models using geodetic temporal data which can possibly predict the shoreline position is an important task and can significantly contribute to coastal management. The studied area is located at municipality of Matinhos in the Paraná State, Brazil. The temporal shoreline used to test the [...] prediction model is respectively from analog photogrammetric data, related to the years 1954, 1963, 1980, 1991 and 1997, and GPS (Global Position System) geodetic surveys for 2001, 2002, 2005 and 2008 (as control). Two different tests with artificial neural network were organized setting the parameters like: architecture, number of neuron in hidden layers and the training algorithms. Comparing the residuals between the prediction to the shoreline of control, the best statistical results show the MAPE (Mean Absolute Percentage Error) is 0,28% using the Elman partially recurrent network with quasi-Newton training function and 0,46% using the neural network multilayer perceptron with Bayesian regulation training function.

Rodrigo Mikosz, Gonçalves; Leandro dos Santos, Coelho; Claudia Pereira, Krueger; Bernhard, Heck.

2010-09-01

383

Scientific Electronic Library Online (English)

Full Text Available SciELO Brazil | Language: Portuguese Abstract in portuguese A estimativa da evapotranspiração por métodos indiretos propicia, de modo facilitado, a geração de dados para o planejamento de sistemas de irrigação e aplicação de modelos meteorológicos e hidrológicos, ambos, úteis na gestão de bacias hidrográficas. O objetivo deste trabalho foi elaborar uma Rede [...] Neural Artificial (RNA) para estimar a evapotranspiração de referência (Eto) em função de dados diários de temperatura do ar. A RNA, do tipo FeedForward Multilayer Perceptron, foi treinada tomando-se por referência a Eto diária obtida pelo método de Penman-Monteith. Nas camadas intermediárias e de saída foram utilizadas funções de ativação do tipo tan-sigmóide e lineares, respectivamente. Os valores de Eto gerados pela RNA foram comparados com os obtidos pelos métodos de Blanney-Criddle e Hargreaves considerando meses referentes às quatro estações do ano. Em relação aos outros métodos analisados, os resultados obtidos a partir da RNA foram mais próximos ao método padrão Penman-Monteith. Assim, o desempenho da RNA desenvolvida foi satisfatório, podendo-se considerá-la como integrante do conjunto de métodos indiretos para estimativa da evapotranspiração, além de representar uma diminuição dos custos de aquisição de dados para estimativa desta variável. Abstract in english The estimation of evapotranspiration by indirect methods provides synthetic data for planning irrigation systems and application on meteorological and hydrological models, both useful in watershed management. The objective of this study was to develop an Artificial Neural Network (ANN) to estimate t [...] he reference evapotranspiration (Eto) based on daily air temperature data. The ANN model of Feedforward Multilayer Perceptron type, was trained using as a reference the daily Eto obtained by the Penman-Monteith method. In the intermediate and output layers were used activation functions like tan-sigmoid and linear, respectively. Eto values generated by ANN were compared with those obtained by the methods of Blanney-Criddle and Hargreaves considering the months of the four seasons. Comparing to the other analyzed methods, the results obtained from the ANN were closer to the standard Penman-Monteith method. Thus, the performance of the developed ANN was satisfactory, and the ANN model can be considered as one indirect method for estimating evapotranspiration and allows a cost reduction on data acquisition to estimate this variable.

Teodorico, Alves Sobrinho; Dulce Buchala Bicca, Rodrigues; Paulo Tarso Sanches de, Oliveira; Lais Cristina Soares, Rebucci; Caroline Alvarenga, Pertussatti.

384

Directory of Open Access Journals (Sweden)

Full Text Available Abstract Background Dementia and cognitive impairment associated with aging are a major medical and social concern. Neuropsychological testing is a key element in the diagnostic procedures of Mild Cognitive Impairment (MCI, but has presently a limited value in the prediction of progression to dementia. We advance the hypothesis that newer statistical classification methods derived from data mining and machine learning methods like Neural Networks, Support Vector Machines and Random Forests can improve accuracy, sensitivity and specificity of predictions obtained from neuropsychological testing. Seven non parametric classifiers derived from data mining methods (Multilayer Perceptrons Neural Networks, Radial Basis Function Neural Networks, Support Vector Machines, CART, CHAID and QUEST Classification Trees and Random Forests were compared to three traditional classifiers (Linear Discriminant Analysis, Quadratic Discriminant Analysis and Logistic Regression in terms of overall classification accuracy, specificity, sensitivity, Area under the ROC curve and Press'Q. Model predictors were 10 neuropsychological tests currently used in the diagnosis of dementia. Statistical distributions of classification parameters obtained from a 5-fold cross-validation were compared using the Friedman's nonparametric test. Results Press' Q test showed that all classifiers performed better than chance alone (p Conclusions When taking into account sensitivity, specificity and overall classification accuracy Random Forests and Linear Discriminant analysis rank first among all the classifiers tested in prediction of dementia using several neuropsychological tests. These methods may be used to improve accuracy, sensitivity and specificity of Dementia predictions from neuropsychological testing.

Santana Isabel

2011-08-01

385

Energy Technology Data Exchange (ETDEWEB)

This paper describes a practical introduction to the use of Artificial Neural Networks. Artificial Neural Nets are often used as an alternative to the traditional symbolic manipulation and first order logic used in Artificial Intelligence, due the high degree of difficulty to solve problems that can not be handled by programmers using algorithmic strategies. As a particular case of Neural Net a Multilayer Perception developed by programming in C language on OS9 real time operating system is presented. A detailed description about the program structure and practical use are included. Finally, several application examples that have been treated with the tool are presented, and some suggestions about hardware implementations. (Author) 15 refs.

Labrador, I.; Carrasco, R.; Martinez, L.

1996-07-01

386

International Nuclear Information System (INIS)

This paper describes a practical introduction to the use of Artificial Neural Networks. Artificial Neural Nets are often used as an alternative to the traditional symbolic manipulation and first order logic used in Artificial Intelligence, due the high degree of difficulty to solve problems that can not be handled by programmers using algorithmic strategies. As a particular case of Neural Net a Multilayer Perception developed by programming in C language on OS9 real time operating system is presented. A detailed description about the program structure and practical use are included. Finally, several application examples that have been treated with the tool are presented, and some suggestions about hardware implementations. (Author) 15 refs

387

Directory of Open Access Journals (Sweden)

Full Text Available Neurological diseases like Alzheimer, epilepsy, Parkinson's disease, multiple sclerosis and other dementias influence the lives of patients, their families and society. Parkinson's disease (PD is a neurodegenerative disease that occurs due to loss of dopamine, a neurotransmitter and slow destruction of neurons. Brain area affected by progressive destruction of neurons is responsible for controlling movements, and patients with PD reveal rigid and uncontrollable gestures, postural instability, small handwriting and tremor. Commercial activity-promoting gaming systems such as the Nintendo Wii and Xbox Kinect can be used as tools for tremor, gait or other biomedical signals acquisitions. They also can aid for rehabilitation in clinical settings. This paper emphasizes the use of intelligent optical sensors or accelerometers in biomedical signal acquisition, and of the specific nonlinear dynamics parameters or fuzzy logic in Parkinson's disease tremor analysis. Nowadays, there is no screening test for early detection of PD. So, we investigated a method to predict PD, based on the image processing of the handwriting belonging to a candidate of PD. For classification and discrimination between healthy people and PD people we used Artificial Neural Networks (Radial Basis Function - RBF and Multilayer Perceptron - MLP and an Adaptive Neuro-Fuzzy Classifier (ANFC. In general, the results may be expressed as a prognostic (risk degree to contact PD.

GEMAN, O.

2014-02-01

388

Aspects of artificial neural networks - with applications in high energy physics

International Nuclear Information System (INIS)

Different aspects of artificial neural networks are studied and discussed. They are demonstrated to be powerful general purpose algorithms, applicable to many different problem areas like pattern recognition, function fitting and prediction. Multi-layer perceptron (MPL) models are shown to out perform previous standard approaches on both off-line and on-line analysis tasks in high energy physics, like quark flavour tagging and mass reconstruction, as well as being powerful tools for prediction tasks. It is also demonstrated how a self-organizing network can be employed to extract information from data, for instance to track down origins of unexpected model discrepancies. Furthermore, it is proved that the MPL is more efficient than the learning vector quantization technique on classification problems, by producing smoother discrimination surfaces, and that an MPL network should be trained with a noisy updating schedule if the Hessian is ill-conditioned - A result that is especially important for MPL network with more than just one hidden layer. 81 refs, 6 figs

389

Directory of Open Access Journals (Sweden)

Full Text Available Considering the significance of the Sodium Adsorption Ratio (SAR for growing plants, its prediction is essential for water quality management for irrigation. The SAR prediction in Chelghazy River in Kurdistan, northwest of Iran, using an Artificial Neural Network (ANN was studied. The study applied the Multilayer Perceptron (MLP of the ANN to average monthly data, which was collected by the water authority of the Kurdistan province for the period of 1998-2009. The input parameters of the MLP network was pH, discharge, sulfate, sodium, calcium, chloride, magnesium and bicarbonate, and output was predictive of the SAR. The results showed a correlation coefficient 0.976 between actual and predicted SAR, which means the accuracy of the model is acceptable. The model uses the input parameters to predict the SAR at the same month. The sensitivity analysis indicated the prediction of the SAR was affected by merely pH and calcium. As a whole, the MLP of the ANN may be applicable for prediction of the SAR which is necessary parameter ration for agriculture.

Gholamreza Asadollahfardi

2013-08-01

390

Developing machine learning and soft computing techniques has provided many opportunities for researchers to establish new analytical methods in different areas of science. The objective of this study is to investigate the potential of two types of intelligent learning methods, artificial neural networks and neuro-fuzzy systems, in order to estimate breeding values (EBV) of Iranian dairy cattle. Initially, the breeding values of lactating Holstein cows for milk and fat yield were estimated using conventional best linear unbiased prediction (BLUP) with an animal model. Once that was established, a multilayer perceptron was used to build ANN to predict breeding values from the performance data of selection candidates. Subsequently, fuzzy logic was used to form an NFS, a hybrid intelligent system that was implemented via a local linear model tree algorithm. For milk yield the correlations between EBV and EBV predicted by the ANN and NFS were 0.92 and 0.93, respectively. Corresponding correlations for fat yield were 0.93 and 0.93, respectively. Correlations between multitrait predictions of EBVs for milk and fat yield when predicted simultaneously by ANN were 0.93 and 0.93, respectively, whereas corresponding correlations with reference EBV for multitrait NFS were 0.94 and 0.95, respectively, for milk and fat production. PMID:22991575

Shahinfar, Saleh; Mehrabani-Yeganeh, Hassan; Lucas, Caro; Kalhor, Ahmad; Kazemian, Majid; Weigel, Kent A

2012-01-01

391

NEURAL NETWORKS FOR THE SIMULATION OF MICROCLIMATIC PARAMETERS IN DAIRY HOUSES

Directory of Open Access Journals (Sweden)

Full Text Available The aim of the present paper is to study natural ventilation in a dairy house by means of a parametric analysis relating wind speed and direction to the air flows through the ridge vent of the building. This analysis was carried out by means of an artificial neural network (ANN which capability in modelling and simulating some climatic parameters inside a dairy