Error Back-propagation Artificial Neural Network Research Articles

Prediction of Spatial Peak Overpressure Profile of Air Blast Shocks Using Multiple Linear Regression and Artificial Neural Network

The shock peak overpressure yielded by energetic materials in the near-earth explosion causes damage to people and property. In the literature, several works have reported the effectiveness in obtaining the peak overpressure in the near field by numerical simulation, real tests, and empirical calculations. However, several issues are associated with the above methods, such as the time-consuming of a complex simulation, the difficulties in placing sensors in an obstacle block site, and the limited scope of applications for empirical equations. At present, there is no single model that can rapidly predict the spatial distribution of shock peak overpressure with ground reflection. In this regard, this paper aims to solve the problem of shock propagation in a large-scale space and obtain a model that can predict a wide range of spatial distribution of peak overpressure in the near-earth air blast simply and quickly. This work uses AUTODYN-2D to obtain the peak overpressure of the cylindrical charge explosion in the air near earth. The peak overpressure from the numerical models agrees with the empirical equations and the real blast results. On this basis, the multiple linear regression (MLR) and the optimal error back-propagation artificial neural network (ANN) models are tried and constructed with three geometric parameters as the input under a typical environment. The input parameters are the straight line scaled distance (SLSD), the vertical scaled distance (VSD), and the scaled height of burst (SHOB), respectively. The prediction results of the MLR and ANN models are verified and compared on the test set. Results show that the ANN model with the geometric parameters as the input is better than the MLR model in predicting the air blast peak overpressure. The former takes no more than 0.2s to compute 1441 sets of inputs, which provides a quick reference for predicting spatial peak overpressure in a large-scale shock wave of the air blast.

Assessment of maize seed vigor under saline-alkali and drought stress based on low field nuclear magnetic resonance

To detect maize seed vigor under salt-alkaline and drought stress conditions, transverse relaxation time, physiological index, and electron microscopy images of germinating seeds were studied under different stress conditions. The results showed that the water in germinating maize seeds exist in bound water (T21), semi-bound water (T22), and free water (T23), in addition parabolic water (T24). The Pearson correlation analysis was performed with T2 relaxation parameters and seed vigor parameters, and this analysis yield nine optimized parameters. To predict vigor levels under different stress conditions, an error backpropagation artificial neural network model was developed, wherein the T2 chirality parameter was used as the input value, and the seed germination indices of different stress levels were used as the output values. The model could predict the environmental stress level of maize seed growth. The optimized parameter set showed a prediction accuracy of 92.50%, thus outperforming the T2 relaxation information model without parameter optimization (75.01%). The proposed method can collect data during the germination of maize seeds without any interference and achieve large-scale prediction of seed development status by small sample collection. With the stress environment was aggravated, the physiological structure of seed cells was changed, and the cell structure was destroyed and the ability of water absorption was disappeared. This analysis provides theoretical support and a reference basis for maize planting and production.

Preparation of Mesoporous Silica Nanosphere-Doped Color-Sensitive Materials and Application in Monitoring the TVB-N of Oysters.

In this work, a new colorimetric sensor based on mesoporous silica nanosphere-modified color-sensitive materials was established for application in monitoring the total volatile basic nitrogen (TVB-N) of oysters. Firstly, mesoporous silica nanospheres (MSNs) were synthesized based on the improved Stober method, then the color-sensitive materials were doped with MSNs. The “before image” and the “after image” of the colorimetric senor array, which was composed of nanocolorimetric-sensitive materials by a charge-coupled device (CCD) camera were then collected. The different values of the before and after image were analyzed by principal component analysis (PCA). Moreover, the error back-propagation artificial neural network (BP-ANN) was used to quantitatively predict the TVB-N values of the oysters. The correlation coefficient of the colorimetric sensor array after being doped with MSNs was greatly improved; the Rc and Rp of BP-ANN were 0.9971 and 0.9628, respectively when the principal components (PCs) were 10. Finally, a paired sample t-test was used to verify the accuracy and applicability of the BP-ANN model. The result shows that the colorimetric-sensitive materials doped with MSNs could improve the sensitivity of the colorimetric sensor array, and this research provides a fast and accurate method to detect the TVB-N values in oysters.

Classifying neural networks and methods of their illogical behaviour revealing

In this article a feedforward error backpropagation artificial neural network is investigated and the analysis of its illogical behaviour is presented. The problem of illogical behavior arises in various models of artificial neural networks. In the presented work a classifying artificial neural network (CANN) is considered and several learning algorithms were implemented and compared. CANN was designed for automatic differentiaition of cyanobacterial strains during environmental monitoring and some of trained networks demonstrated illogical behavior in further testing. Several original techniques were elaborated for estimation of the quality and accuracy of classification in addition to the traditional ones. Novel visualization methods were suggested for classification and generalization results representation.

A quantitative relationship between Tgs and chain segment structures of polystyrenes

Abstract The glass transition temperature (Tg) is a fundamental characteristic of an amorphous polymer. A quantitative structure-property relationship (QSPR) based on error back-propagation artificial neural network (ANN) was constructed to predict Tgs of 107 polystyrenes. Stepwise multiple linear regression (MLR) analysis was adopted to select an optimal subset of molecular descriptors. The chain segments (or motion units) of polymer backbones with 20 carbons in length (10 repeating units) were used to calculate these molecular descriptors reflecting polymer structures. The relative optimal conditions of ANN were obtained by adjusting various network paramters by trial-and-error. Compared to the model already published in the literature, the optimal ANN model with [4-7-1] network structure in this paper is accurate and acceptable, although our model has more samples in the test set. The results demonstrate the feasibility and powerful ability of the chain segment structures as representative of polymers for developing Tg models of polystyrenes.

Microstructure-Tensile Properties Correlation for the Ti-6Al-4V Titanium Alloy

Finding the quantitative microstructure-tensile properties correlations is the key to achieve performance optimization for various materials. However, it is extremely difficult due to their non-linear and highly interactive interrelations. In the present investigation, the lamellar microstructure features-tensile properties correlations of the Ti-6Al-4V alloy are studied using an error back-propagation artificial neural network (ANN-BP) model. Forty-eight thermomechanical treatments were conducted to prepare the Ti-6Al-4V alloy with different lamellar microstructure features. In the proposed model, the input variables are microstructure features including the α platelet thickness, colony size, and β grain size, which were extracted using Image Pro Plus software. The output variables are the tensile properties, including ultimate tensile strength, yield strength, elongation, and reduction of area. Fourteen hidden-layer neurons which can make ANN-BP model present the most excellent performance were applied. The training results show that all the relative errors between the predicted and experimental values are within 6%, which means that the trained ANN-BP model is capable of providing precise prediction of the tensile properties for Ti-6Al-4V alloy. Based on the corresponding relations between the tensile properties predicted by ANN-BP model and the lamellar microstructure features, it can be found that the yield strength decreases with increasing α platelet thickness continuously. However, the α platelet thickness exerts influence on the elongation in a more complicated way. In addition, for a given α platelet thickness, the yield strength and the elongation both increase with decreasing β grain size and colony size. In general, the β grain size and colony size play a more important role in affecting the tensile properties of Ti-6Al-4V alloy than the α platelet thickness.

High Accuracy Arabic Handwritten Characters Recognition Using Error Back Propagation Artificial Neural Networks

This manuscript considers a new architecture to handwritten characters recognition based on simulation of the behavior of one type of artificial neural network, called the Error Back Propagation Artificial Neural Network (EBPANN). We present an overview of our neural network to be optimized and tested on 12 offline isolated Arabic handwritten characters (???????????????? ?,?,???) because the similarity of some Arabic characters and the location of the points in the character. Accuracy of 93.61% is achieved using EBPANN which is the highest accuracy achieved during Offline Handwritten Arabic Character Recognition. It is noted that the EBPANN in general generates an optimized comparison between the input samples and database samples which improves the final recognition rate. Experimental results show that the EBPANN is convergent and more accurate in solutions that minimize the error recognition rate.

Consumer credit scoring models with limited data

In this paper we design the neural network consumer credit scoring models for financial institutions where data usually used in previous research are not available. We use extensive primarily accounting data set on transactions and account balances of clients available in each financial institution. As many of these numerous variables are correlated and have very questionable information content, we considered the issue of variable selection and the selection of training and testing sub-sets crucial in developing efficient scoring models. We used a genetic algorithm for variable selection. In dividing performing and nonperforming loans into training and testing sub-sets we replicated the distribution on Kohonen artificial neural network, however, when evaluating the efficiency of models, we used k-fold cross-validation. We developed consumer credit scoring models with error back-propagation artificial neural networks and checked their efficiency against models developed with logistic regression. Considering the dataset of questionable information content, the results were surprisingly good and one of the error back-propagation artificial neural network models has shown the best results. We showed that our variable selection method is well suited for the addressed problem.

Classification of EMG signals using wavelet neural network

An accurate and computationally efficient means of classifying electromyographic (EMG) signal patterns has been the subject of considerable research effort in recent years. Quantitative analysis of EMG signals provides an important source of information for the diagnosis of neuromuscular disorders. Following the recent development of computer-aided EMG equipment, different methodologies in the time domain and frequency domain have been followed for quantitative analysis. In this study, feedforward error backpropagation artificial neural networks (FEBANN) and wavelet neural networks (WNN) based classifiers were developed and compared in relation to their accuracy in classification of EMG signals. In these methods, we used an autoregressive (AR) model of EMG signals as an input to classification system. A total of 1200 MUPs obtained from 7 normal subjects, 7 subjects suffering from myopathy and 13 subjects suffering from neurogenic disease were analyzed. The success rate for the WNN technique was 90.7% and for the FEBANN technique 88%. The comparisons between the developed classifiers were primarily based on a number of scalar performance measures pertaining to the classification. The WNN-based classifier outperformed the FEBANN counterpart. The proposed WNN classification may support expert decisions and add weight to EMG differential diagnosis.

Calculation of polyamides melting point by quantum-chemical method and BP artificial neural networks

Quantitative structure-property relationships (QSPR) for the melting point of the polyamides have been determined. All descriptors were calculated from molecular structures at the B3LYP/6-31G(d) level and a QSPR model was generated by multiple linear regression (MLR). The important molecular descriptors for polyamide melting-point temperatures (T (m)) are the number of benzene rings in the backbone chain, the proportion of methylene and acylamino in the backbone chain, the total molecular energy and the atomic charge for the oxygen atom in the acylamino group. The MLR determination coefficient (r2) and the standard error of estimation for the model are 0.865 and 21.34 K, respectively. In addition to the nonlinear regression technique, error back-propagation artificial neural networks (BPANN) was used to study the relationships between molecular structures and melting-point temperatures. It is concluded that melting-point temperatures for polyamides can be described by molecular chain rigidity and interchain attractive interactions. The more accurate predicted results were obtained from BPANN. Figure Experimental vs calculated Tm using BPANN. [Figure: see text].

Prediction of polyamide properties using quantum-chemical methods and BP artificial neural networks

Quantitative structure -property relationships (QSPR) for glass translation temperatures (T (g)), density (rho) and indices of refraction (n) of the polyamides have been determined. All descriptors are calculated from molecular structures at the B3LYP/6-31G(d) level. These QSPR models are generated by two methods: multiple linear regression (MLR) and error back-propagation artificial neural networks (BPANN). The model obtained by MLR is used for the calculations of T (g) (R (training)=0.9074, SD(training)=22.4687, R (test)=0.8898, SD(test)=23.2417), rho (R (training)=0.9474, SD(training)=0.0422, R (test)=0.8928, SD (test)=0.0422), n (R (training)=0.9298, SD(training)=0.0204, R (test)=0.9095, SD(test)=0.0274). The model obtained by BPANN is used for the calculations of T (g) (R (training)=0.9273, SD(training)=14.8988, R (test)=0.8989, SD(test)=16.4396), rho (R (training)=0.9523, SD(training)=0.0466, R (test)=0.9014, SD(test)=0.0512), n (R (training)=0.9401, SD(training)=0.0131, R (test)=0.9445, SD(test)=0.0179). These results demonstrate that the MLR and BPANN methods can be used to predict T (g), rho and n. The more accurate predicted results are obtained from BPANN. Figure: Experimental vs. calculated n with cross-validation method (BPANN) for the training set of 53 polyamides and the test set of 14 polyamides. Figure Experimental vs. calculated n with cross-validation method (BPANN) for the training set of 53 polyamides and the test set of 14 polyamides. [Figure: see text].

ANN-based sediment yield models for Vamsadhara river basin (India)

Most universally accepted feed-forward error back-propagation artificial neural network models, supported by batch- and pattern-learning, daily, weekly, ten-daily and monthly sediment yield were developed for the Vamsadhara River basin of India. The fast gradient descent optimisation technique improved with variable learning rate (α) and momentum term (β) was used for optimisation. In the process of optimisation and updating of weights, criteria adopted to terminate the process of learning was selected as a per-decided high number of iteration and the other is the generalisation of model through crossvalidation. In all cases of model formulation, the data were normalised with the maximum value of the variable of the series individually. The pattern-learned models were found superior to batch-learned models. High numbers of iterations adopted for model development were found to reduce the value of the objective function, but with model's over-learning and that is reflected? Unclear what is meant by an increase and decrease of the performance in calibration and cross-validation, respectively. The generalised pattern- learned models for different time scales were compared with linear transfer function models and it was found that the pattern-learned models developed with generalisation through cross-validation were superior in general, except weekly for the study area. Key words: back propagation artificial neural network, sediment yield modelling, generalised modelling. Water SA Vol.31(1) 2005: 95-100

Wavelet neural network classification of EEG signals by using AR model with MLE preprocessing

Since EEG is one of the most important sources of information in therapy of epilepsy, several researchers tried to address the issue of decision support for such a data. In this paper, we introduce two fundamentally different approaches for designing classification models (classifiers); the traditional statistical method based on logistic regression and the emerging computationally powerful techniques based on artificial neural networks (ANNs). Logistic regression as well as feedforward error backpropagation artificial neural networks (FEBANN) and wavelet neural networks (WNN) based classifiers were developed and compared in relation to their accuracy in classification of EEG signals. In these methods we used FFT and autoregressive (AR) model by using maximum likelihood estimation (MLE) of EEG signals as an input to classification system with two discrete outputs: epileptic seizure or nonepileptic seizure. By identifying features in the signal we want to provide an automatic system that will support a physician in the diagnosing process. By applying AR with MLE in connection with WNN, we obtained novel and reliable classifier architecture. The network is constructed by the error backpropagation neural network using Morlet mother wavelet basic function as node activation function. The comparisons between the developed classifiers were primarily based on analysis of the receiver operating characteristic (ROC) curves as well as a number of scalar performance measures pertaining to the classification. The WNN-based classifier outperformed the FEBANN and logistic regression based counterpart. Within the same group, the WNN-based classifier was more accurate than the FEBANN-based classifier, and the logistic regression-based classifier.

Top-of-atmosphere flux retrievals from CERES using artificial neural networks

The Clouds and the Earth's Radiant Energy System (CERES) instruments on the Terra spacecraft provide accurate shortwave (SW), longwave (LW) and window (WN) region top-of-atmosphere (TOA) radiance measurements from which TOA radiative flux values are obtained by applying Angular Distribution Models (ADMs). These models are developed empirically as functions of the surface and cloud properties provided by coincident high-resolution imager measurements over CERES field-of-view. However, approximately 5.6% of the CERES/ Terra footprints lack sufficient imager information for a reliable scene identification. To avoid any systematic biases in regional mean radiative fluxes, it is important to provide TOA fluxes for these footprints. For this purpose, we apply a feedforward error-backpropagation Artificial Neural Network (ANN) technique to reproduce CERES/ Terra ADMs relying only on CERES measurements. All-sky ANN-based angular distribution models are developed for 10 surface types separately for shortwave, longwave and window TOA flux retrievals. To optimize the ANN performance, we use a partially connected first hidden neuron layer and compact training sets with reduced data noise. We demonstrate the performance of the ANN-based ADMs by comparing TOA fluxes inferred from ANN and CERES anisotropic factors. The global annual average bias in ANN-derived fluxes relative to CERES is less than 0.5% for all ANN scene types. The maximum bias occurs over sea ice and permanent snow surfaces. For all surface types, instantaneous ANN-derived TOA fluxes are self-consistent in viewing zenith angle to within 9% for shortwave, 3.5% and 3% longwave daytime and nighttime, respectively.

Application of Remote Sensing and GIS to Model Mountainous Rivers

The purpose of this study is to develop a semidistributed, physically based hydrologic model ~SDPB - HM! for mountainous watersheds areas. Most of the model's required parameters were acquired using remote sensing and digital terrain elevation data. A three-stage computer classifier model was built based on the error back-propagation artificial neural network approach~EBPANN! and refined to classify the land cover types of the study area. In addition, the reflection properties of the land cover types were used to improve the technique of estimating the net radiation in Morton's evapotranspiration model. A procedure was proposed for discretizing the watershed areas, aimed to increase the homogeneity and minimize the calculation time. The SDPB - HM was applied to the Albert River Basin in the Rocky Mountains in British Columbia, Canada. The Albert River meteorological data from October 1986 to September 1987 was used to calibrate the model parameters. In addition, the SDPB - HM was validated using the meteorological data of a case study from October 1987 to September 1988 and from October 1988 to September 1989. Comparison between the simulated and the observed flow at the outlet of river showed a good agreement during these periods.

Comparison of logistic regression and neural network-based classifiers for bacterial growth

In this paper, we introduce two fundamentally different approaches for designing classification models (classifiers); the traditional statistical method based on logistic regression and the emerging computationally powerful techniques based on artificial neural networks (ANNs). Linear and nonlinear logistic regression as well as feedforward error backpropagation artificial neural networks (FEBANN) and probabilistic neural networks (PNN) based classifiers were developed and compared in relation to their accuracy in classification of bacterial growth/no–growth data pertaining to pathogenic Escherichia coli R31 as affected by temperature and water activity. The comparisons between the four developed classifiers were primarily based on analysis of the receiver operating characteristic (ROC) curves as well as a number of scalar performance measures pertaining to the classification contingency matrices. The ANN-based classifiers outperformed the logistic regression based counterparts. Within the same group, the PNN-based classifier was more accurate than the FEBANN-based classifier, and the nonlinear logistic regression-based classifier was more accurate than the linear one. The optimal PNN-based classifier was a perfect classifier with 100% growth detection accuracy and zero false alarm rate. The advantages and limitations pertaining to the development of the various classifiers were discussed.

Predicting nucleic acid torsion angle values using artificial neural networks.

By means of an error back-propagation artificial neural network, a new method to predict the torsion angles, chi, zeta and alpha from torsion angles delta, epsilon, beta and gamma for nucleic acid dinucleotides is introduced. To build a model, training sets and test sets of 163 and 81 dinucleotides, respectively, with known crystal structures, were assembled. With 7 hidden units in a three-layered network a model with good predictive ability is constructed. About 70 to 80% of the residuals for predicted torsion angles are smaller than 10 degrees. This means that such a model can be used to construct trial structures for conformational analysis that can be refined further. Moreover, when reasonable estimates for delta, epsilon, beta and gamma are extracted from COSY experiments, this procedure can easily be extended to predict torsion angles for structures in solution.