Backpropagation Training Algorithm Research Articles

Application of artificial neural network for the prediction of stock market returns: The case of the Japanese stock market

Accurate prediction of stock market returns is a very challenging task because of the highly nonlinear nature of the financial time series. In this study, we apply an artificial neural network (ANN) that can map any nonlinear function without a prior assumption to predict the return of the Japanese Nikkei 225 index. (1) To improve the effectiveness of prediction algorithms, we propose a new set of input variables for ANN models. (2) To verify the prediction ability of the selected input variables, we predict returns for the Nikkei 225 index using the classical back propagation (BP) learning algorithm. (3) Global search techniques, i.e., a genetic algorithm (GA) and simulated annealing (SA), are employed to improve the prediction accuracy of the ANN and overcome the local convergence problem of the BP algorithm. It is observed through empirical experiments that the selected input variables were effective to predict stock market returns. A hybrid approach based on GA and SA improve prediction accuracy significantly and outperform the traditional BP training algorithm.

Model of Cholera Forecasting Using Artificial Neural Network in Chabahar City, Iran

Background: Cholera as an endemic disease remains a health issue in Iran despite decrease in incidence. Since forecasting epidemic diseases provides appropriate preventive actions in disease spread, different forecasting methods including artificial neural networks have been developed to study parameters involved in incidence and spread of epidemic diseases such as cholera. Objectives: In this study, cholera in rural area of Chabahar, Iran was investigated to achieve a proper forecasting model. Materials and Methods: Data of cholera was gathered from 465 villages, of which 104 reported cholera during ten years period of study. Logistic regression modeling and correlate bivariate were used to determine risk factors and achieve possible predictive model one-hidden-layer perception neural network with backpropagation training algorithm and the sigmoid activation function was trained and tested between the two groups of infected and non-infected villages after preprocessing. For determining validity of prediction, the ROC diagram was used. The study variables included climate conditions and geographical parameters. Results: After determining significant variables of cholera incidence, the described artificial neural network model was capable of forecasting cholera event among villages of test group with accuracy up to 80%. The highest accuracy was achieved when model was trained with variables that were significant in statistical analysis describing that the two methods confirm the result of each other. Conclusions: Application of artificial neural networking assists forecasting cholera for adopting protective measures. For a more accurate prediction, comprehensive information is required including data on hygienic, social and demographic parameters.

A Q-backpropagated time delay neural network for diagnosing severity of gait disturbances in Parkinson’s disease

Parkinson’s disease (PD) is a movement disorder that affects the patient’s nervous system and health-care applications mostly uses wearable sensors to collect these data. Since these sensors generate time stamped data, analyzing gait disturbances in PD becomes challenging task. The objective of this paper is to develop an effective clinical decision-making system (CDMS) that aids the physician in diagnosing the severity of gait disturbances in PD affected patients. This paper presents a Q-backpropagated time delay neural network (Q-BTDNN) classifier that builds a temporal classification model, which performs the task of classification and prediction in CDMS. The proposed Q-learning induced backpropagation (Q-BP) training algorithm trains the Q-BTDNN by generating a reinforced error signal. The network’s weights are adjusted through backpropagating the generated error signal. For experimentation, the proposed work uses a PD gait database, which contains gait measures collected through wearable sensors from three different PD research studies. The experimental result proves the efficiency of Q-BP in terms of its improved classification accuracy of 91.49%, 92.19% and 90.91% with three datasets accordingly compared to other neural network training algorithms.

Classification of pepper seeds using machine vision based on neural network

Pepper is widely planted and used all over the world as fresh vegetable and spice. Genetic and morphological information of pepper are stored through seeds. Determination of seed variety is crucial for correctly identifying genetic materials. Pepper varieties cannot be easily classified even by an expert eye due to the very small size of seeds and visual similarities. Hence, more advanced technologies are required to determine the variety of a pepper seed. A classification method was proposed to discriminate pepper seed based on neural networks and computer vision. Image acquisition was conducted using an office scanner at a resolution of 1200 dpi. Image features representing color, shape, and texture were extracted and used to classify pepper seeds. By calculating features from different color components, a feature database was constructed. Effective features were selected using sequential feature selection with different criterion functions. As a result of the feature selection procedure, the number of the features was significantly reduced from 257 to 10. Cross validation rules were applied to obtain a reliable classification model by preventing overfitting. Different numbers of neurons in the hidden layer and various training algorithms were investigated to determine the best multilayer perceptron model. The best classification performance was obtained using 30 neurons in the hidden layer of the network. With this network, an accuracy rate of 84.94% was achieved using the sequential feature selection and the training algorithm of resilient back propagation in classifying eight pepper seed varieties. Keywords: pepper seed, neural networks, variety classification, computer vision DOI: 10.3965/j.ijabe.20160901.1790 Citation: Kurtulmus F, Alibas I, Kavdir I. Classification of pepper seeds using neural network. Int J Agric & Biol Eng, 2016; 9(1): 51－62.

Kinetics, equilibrium isotherm and neural network modeling studies for the sorption of hexavalent chromium from aqueous solution by quartz/feldspar/wollastonite

A three layer feed forward artificial neural network (ANN) with back propagation training algorithm was developed to model the adsorption process of Cr(vi) in aqueous solution using riverbed sand containing quartz/feldspar/wollastonite (QFW) as adsorbent.

Digital image classification with the help of artificial neural network by simple histogram.

Background:Visual image classification is a great challenge to the cytopathologist in routine day-to-day work. Artificial neural network (ANN) may be helpful in this matter.Aims and Objectives:In this study, we have tried to classify digital images of malignant and benign cells in effusion cytology smear with the help of simple histogram data and ANN.Materials and Methods:A total of 404 digital images consisting of 168 benign cells and 236 malignant cells were selected for this study. The simple histogram data was extracted from these digital images and an ANN was constructed with the help of Neurointelligence software [Alyuda Neurointelligence 2.2 (577), Cupertino, California, USA]. The network architecture was 6-3-1. The images were classified as training set (281), validation set (63), and test set (60). The on-line backpropagation training algorithm was used for this study.Result:A total of 10,000 iterations were done to train the ANN system with the speed of 609.81/s. After the adequate training of this ANN model, the system was able to identify all 34 malignant cell images and 24 out of 26 benign cells.Conclusion:The ANN model can be used for the identification of the individual malignant cells with the help of simple histogram data. This study will be helpful in the future to identify malignant cells in unknown situations.

Application of an artificial neural network for a direct estimation of atmospheric instability from a next-generation imager

Atmospheric instability information derived from satellites plays an important role in short-term weather forecasting, especially the forecasting of severe convective storms. For the next generation of weather satellites for Korea’s multi-purpose geostationary satellite program, a new imaging instrument has been developed. Although this imaging instrument is not designed to perform full sounding missions and its capability is limited, its multi-spectral infrared channels provide information on vertical sounding. To take full advantage of the observation data from the much improved spatiotemporal resolution of the imager, the feasibility of an artificial neural network approach for the derivation of the atmospheric instability is investigated. The multi-layer perceptron model with a feed-forward and back-propagation training algorithm shows quite a sensitive response to the selection of the training dataset and model architecture. Through an extensive performance test with a carefully selected training dataset of 7197 independent profiles, the model architectures are selected to be 12, 5000, and 0.3 for the number of hidden nodes, number of epochs, and learning rate, respectively. The selected model gives a mean absolute error, RMSE, and correlation coefficient of 330 J kg-1, 420 J kg-1, and 0.9, respectively. The feasibility is further demonstrated via application of the model to real observation data from a similar instrument that has comparable observation channels with the planned imager.

Prediction of the Liquid Vapor Pressure Using the Artificial Neural Network-Group Contribution Method

In this paper, vapor pressure for pure compounds is estimated using the Artificial Neural Networks and a simple Group Contribution Method (ANN-GCM). For model comprehensiveness, materials were chosen from various families. Most of materials are from 12 families. Vapor pressure data of 100 compounds is used to train, validate and test the ANN-GCM model. Vapor pressure data were taken from literature for wide ranges of temperature (68.55-559.15 K). Based on results, the best structure for feed-forward back propagation neural network is Levenberg-Marquardt back propagation training algorithm, logsig transfer function for hidden layer and linear transfer function for output layer. The multiplayer network model consists of temperature, acentric factor, critical temperature, critical pressure and the structure of molecules as inputs, 10 neurons in the hidden layer and one neuron in the output layer corresponding to vapor pressure. The weights are optimized to minimize error between experimental and calculated data. Results show that optimum neural network architecture is able to predict vapor pressure data with an acceptable level. The trained network predicts the vapor pressure data with average relative deviation percent of 1.18%.

Elman Recurrent Neural Network Application in Adaptive Beamforming of Smart Antenna System

In this paper an artificial Elman Recurrent Neural Network (ERNN) is used for smart antenna adaptive beamforming. Neural network is used to calculate the optimum weights of uniform linear array antenna that steer the radiation pattern of the antenna by directing multiple narrow beams toward the desired users and make nulling in the direction of unwanted users. Two different supervised training algorithms are used to train the ERNN , they are Levenberg Marquardt (LM) algorithm and Resilient Backpropagation (Rprop) algorithm. Uniform linear array is used with five element and the spacing between element equal to half wavelength .The results of ERNN training using LM and Rprop showed that the performance of Neural Network (NN) trained by LM training algorithm is better than Rprop training algorithm ,since it consider the fastest backpropagation training algorithm but it requires more memory than other algorithms.

Time prediction models of grapple skidder HSM 904 using multiple linear regressions (MLR) and adaptive neuro-fuzzy inference system (ANFIS)

The majority of productivity studies in forest operations use the ordinary least square regression in order to calculate time prediction models. Objective of this paper was to compare two methods, the adaptive neuro fuzzy inference system (ANFIS) and the multiple linear regression (MLR) pertaining to the modeling the loading time with the grapple skidder HSM 904. In order to investigate the influence of changes in membership functions (MF), inference system (IS) and optimization methods on the performance of the ANFIS model, four types of MF, two types of IS and two types of optimization methods were applied. Two stepwise (forward and backward) techniques were applied in the development of the MLR models. The data originated from a time study of 35 truckloads including 238 loading cycles with the specific forest machine. The comparison of the various modeling approaches indicated that the generated ANFIS with constant IS, Backpropagation training algorithm, and Gaussian membership function had a greater predictive power (R2 = 0.84) and higher performance than all calculated MLR models (R2 = 0.61). The results also showed that ANFIS could predict with a relatively high accuracy (R2 = 0.74) the loading time by adopting the “number of loaded logs” as the single input variable.

The Knowledge-Based Modeling of Ferroelectric Hysteresis Area: An Application to Forming (1-x)PZT-(x)PZN Hysteresis Database

In this work, Artificial Neural Network was used to model the hysteresis behavior of lead zirconate titanate-lead zinc niobate (Pb(Zr1/2Ti1/2)O3−Pb(Zn1/3Nb2/3)O3 or (1-x)PZT-(x)PZN mixed ferroelectric systems. The hysteresis loops were measured with varying electric filed parameters and the composition x of the mixed ferroelectrics. A knowledge-based technique, i.e. the Artificial Neural Network (ANN), was employed in modeling the hysteresis to construct the database of how field parameters and the mixed composition affect dynamic hysteresis behavior. The input data to the ANN were composition x, field amplitude E0 and field frequency f, where the output data was the hysteresis area. The inputs-outputs were divided into training, validating and testing datasets for the ANN. Multilayer perceptron with back propagation training algorithm was applied in this work. Exhaustive search was used to obtain the best network algorithm that gives minimum error in the training process. With the best network, unseen input datasets were fed into the network to predict hysteresis area. From the results, the predicted and the actual data match very well over an extensive range of field parameters, where the scattering plot between the predicted and the actual area has R-squared greater than 0.99. This therefore indicates ANN capabilities in modeling dynamic-hysteresis phenomena across (1-x)PZT-(x)PZN systems even they have different ratios of structural phases at microscopic level.

Artificial neural network modeling of the effluent quality index for municipal waste water treatment plants using quality variables: south of Tehran waste water treatment plant

Using treated wastewater could be a realistic solution for the scarcity of fresh water in Iran; therefore, effluent quality evaluation is an essential task. In the present study, the effluent quality index (EQI) for south of Tehran municipal wastewater treatment plant was found considering eight quality parameters and sub-index formulae method. Further, with the help of an artificial neural network (ANN), design and applicability of feed-forward, three-layer perceptron neural network model for computing the EQI has been assessed. In this study, ANN modeling is done by ANN toolbox in MATLAB 2013. The modeling efforts showed that the optimal network architecture was 8–7–1 and that the best EQI predictions were associated with the Levenberg–Marquardt back propagation training algorithm and Tansig transfer function. The EQI predictions of this model had significant and very high correlation (R = 0.96, MSE = 0.1) with the measured EQI values. The ANN approach which is used in this article suggested powerful tool to EQI computation and prediction, since it reduced lengthy computations and using various sub-index formula for each value.

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

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

Long Term and Short Term Investment Strategy for Predicting the Performance of BSE using MLP Model

Background/Objectives: Now a day’s stock market has great influence in the daily life of people. The biggest problem of stock market is that it holds large uncertainties which are directly related to their long term and short term investment. The key objective of researchers in this area is to strengthen economy as well as profit maximization on individual share. Methods/Statistical Analysis: This study, adopted a most popular back propagation training algorithm for predicting the performance of Bombay Stock Exchange of India based on long term and short term basis. In this study, 3500 samples of nine different companies are considered are divided 60% for training, 20% for validation and remaining 20% for testing task. In the initial phase the data underwent five phases which includes variable selection, data preprocessing, training dataset, prediction and evaluation. However the research evaluated through MATLAB tool which consists of training performance, error rate, and output prices. After getting the output from various evaluations made on stock dataset. Findings: It is shown by experimental results the method has given more accurate prediction results. Besides, it takes less training time and epochs for prediction task. It is also found that long term investment also has more successive returns over short term investment. Applications/Improvements: In the present study back propagation method is employed for long term and short term prediction. The different experimental results are carried out on stock data sets. It is also observed by study for long span of time small scale and medium scale companies like HCL and Ambuja Cement have better performance compared to large scale companies like TCS. In contrast, short span of time large scale companies like TCS also have better performance.

A novel method for early diagnosis of Alzheimer’s disease based on pseudo Zernike moment from structural MRI

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the most common type of dementia among older people. The number of patients with AD will grow rapidly each year and AD is the fifth leading cause of death for those aged 65 and older. In recent years, one of the main challenges for medical investigators has been the early diagnosis of patients with AD because an early diagnosis can provide greater opportunities for patients to be eligible for more clinical trials and they will have enough time to plan for future, medical and financial decisions. An established risk factor for AD is mild cognitive impairment (MCI) which is described as a transitional state between normal aging and AD patients. Hence an accurate and reliable diagnosis of MCI can be very effective and helpful for early diagnosis of AD. Therefore in this paper we present a novel and efficient method based on pseudo Zernike moments (PZMs) for the diagnosis of MCI individuals from AD and healthy control (HC) groups using structural MRI. The proposed method uses PZMs to extract discriminative information from the MR images of the AD, MCI, and HC groups. Two types of artificial neural networks, which are based on pattern recognition and learning vector quantization (LVQ) networks, were used to classify the information extracted from the MRIs. We worked with 500 MRIs from the database of the Alzheimer’s Disease Neuroimaging Initiative (ADNI 1 1.5T). The 1 slice of 500 MRIs used in this study included 180 AD patients, 172 MCI patients, and 148 HC individuals. We selected 50 percent of the MRIs randomly for use in training the classifiers, 25 percent for validation and we used 25 percent for the testing phase. The technique proposed here yielded the best overall classification results between AD and MCI (accuracy 94.88%, sensitivity 94.18%, and specificity 95.55%), and for pairs of the MCI and HC (accuracy 95.59%, sensitivity 95.89% and specificity 95.34%). These results were achieved using maximum order 30 of PZM and the pattern recognition network with the scaled conjugate gradient (SCG) back-propagation training algorithm as a classifier.

Electrical consumption forecasting in hospital facilities: An application case

The topic of energy efficiency applied to buildings represents one of the key aspects in today's international energy policies. Emissions reduction and the achievement of the targets set by the Kyoto Protocol are becoming a fundamental concern in the work of engineers and technicians operating in the energy management field. Optimal energy management practices need to deal with uncertainties in generation and demand, hence the development of reliable forecasting methods is an important priority area of research in electric energy systems. This paper presents a load forecasting model and the way it was applied to a real case study, to forecast the electrical consumption of the Cellini medical clinic of Turin. The model can be easily integrated into a Building Management System or into a real time monitoring system. The load forecasting is performed through the implementation of an artificial neural network (ANN). The proposed multi-layer perceptron ANN, based on a back propagation training algorithm, is able to take as inputs: loads, data concerning the type of day (e.g. weekday/holiday), time of the day and weather data. In particular, this work focuses on providing a detailed analysis and an innovative formal procedure for the selection of all the ANN parameters.

Development of ANN model for prediction of performance and emission characteristics of hydrogen dual fueled diesel engine with Jatropha Methyl Ester biodiesel blends

The present study investigates the use of Artificial Neural Network modeling for prediction of performance and emission characteristics of a four stroke single cylinder diesel engine with Jatropha Methyl Ester biodiesel blends along with hydrogen in dual fuel mode. ANN model was developed to predict BTE, BSFC, CO, O2, CO2, NOx, HC and EGT based on initial experimental studies by varying load, blends of biodiesel and hydrogen flow rates. Seven training algorithms each with five combinations of trainings functions were investigated. Levenberg-Marquardt backpropagation training algorithm with logarithmic sigmoid and hyperbolic tangent sigmoid transfer function results in best model for prediction of performance and emissions characteristics. The overall regression coefficient, MSE and MAPE for the model developed are 0.99360, 0.0011 and 4.863001% respectively. It is found that the neural networks are good tools for simulation and prediction of dual fueled hydrogen jatropha biodiesel engine.

Evaluation of wavelet performance via an ANN-based electrical conductivity prediction model.

The prediction of water quality parameters plays an important role in water resources and environmental systems. The use of electrical conductivity (EC) as a water quality indicator is one of the important parameters for estimating the amount of mineralization. This study describes the application of artificial neural network (ANN) and wavelet-neural network hybrid (WANN) models to predict the monthly EC of the Asi River at the Demirköprü gauging station, Turkey. In the proposed hybrid WANN model, the discrete wavelet transform (DWT) was linked to the ANN model for EC prediction using a feed-forward back propagation (FFBP) training algorithm. For this purpose, the original time series of monthly EC and discharge (Q) values were decomposed to several sub-time series by DWT, and these sub-time series were then presented to the ANN model as an input dataset to predict the monthly EC. Comparing the values predicted by the models indicated that the performance of the proposed WANN model was better than the conventional ANN model. The correlation of determination (R (2)) were 0.949 and 0.381 for the WANN and ANN models, respectively. The results indicate that the peak EC values predicted by the WANN model are closer to the observed values, and this model simulates the hysteresis phenomena at an acceptable level as well.

Artificial neural network modeling of the water quality index using land use areas as predictors.

This paper describes the design of an artificial neural network (ANN) model to predict the water quality index (WQI) using land use areas as predictors. Ten-year records of land use statistics and water quality data for Kinta River (Malaysia) were employed in the modeling process. The most accurate WQI predictions were obtained with the network architecture 7-23-1; the back propagation training algorithm; and a learning rate of 0.02. The WQI forecasts of this model had significant (p < 0.01), positive, very high correlation (ρs = 0.882) with the measured WQI values. Sensitivity analysis revealed that the relative importance of the land use classes to WQI predictions followed the order: mining > rubber > forest > logging > urban areas > agriculture > oil palm. These findings show that the ANNs are highly reliable means of relating water quality to land use, thus integrating land use development with river water quality management.

Improving the Performance of Back-Propagation Training Algorithm by Using ANN

Improving the Performance of Back-Propagation Training Algorithm by Using ANN