Robust human iris pattern recognition system using neural network approach

In this article, a comparison study of three artificial intelligence (AI) techniques for energy consumption estimation are presented. The models considered are: multilayer perceptron (MLP); radial basis function (RBF) and support vector machine (SVM). The energy consumption is modeled as a function of activity, structural and intensity changes. The models are applied to Canadian industrial manufacturing data from 1990 to 2000. Comparisons were based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Relative Absolute Error (RAE), Root Relative Square Error (RRSE) as well as Simulation Time. The best results were obtained for the Multilayer Perceptron.

Neural network based detection of hard exudates in retinal images

Estimating the Technical Deterioration of Large-panel Residential Buildings Using Artificial Neural Networks

Abstract:Bankruptcy prediction and credit scoring are the two important problems facing financial decision support. The multilayer perceptron (MLP) network has shown its applicability to these problems and its performance is usually superior to those of other traditional statistical models. Support vector machines (SVMs) are the core machine learning techniques and have been used to compare with MLP as the benchmark. However, the performance of SVMs is not fully understood in the literature because an insufficient number of data sets is considered and different kernel functions are used to train the SVMs. In this paper, four public data sets are used. In particular, three different sizes of training and testing data in each of the four data sets are considered (i.e. 3:7, 1:1 and 7:3) in order to examine and fully understand the performance of SVMs. For SVM model construction, the linear, radial basis function and polynomial kernel functions are used to construct the SVMs. Using MLP as the benchmark, the SVM classifier only performs better in one of the four data sets. On the other hand, the prediction results of the MLP and SVM classifiers are not significantly different for the three different sizes of training and testing data.

Performance of modern automated pattern recognition (PR) systems is heavily influenced by accuracy of their feature extraction algorithm. Many papers have demonstrated uses of deep learning techniques in PR, but there is little evidence on using them as feature extractors. Our goal is to contribute to this field and perform a comparative study between classical used methods in feature extraction and deep learning techniques. For that, a biometric recognition system, which is a PR application, is developed and evaluated using a proposed evaluation metric called expected risk probability. In our study, two deeply learned features, based on PCANet and DCTNet deep learning techniques, are used with two biometric modalities that are palmprint and palm-vein. Subsequently, the efficiency of these techniques is compared with various classical feature extraction methods. From the obtained results, we drew our conclusions on a very positive impact of deep learning techniques on overall recognition rate, and thus these techniques significantly outperform the classical techniques.

ABSTRACTEnergy management and reduction of CO2 emission lead to many investigates about energy input–output analyses especially in agricultural sector. The main objectives of this study are to assess the energy use pattern and to select the best method among Cobb–Douglas (CD), multiple linear regressions (MLR), multilayer perceptron (MLP), radial basis function (RBF) and support vector machine (SVM) models to estimate potato output energy in Jiroft city, located in the south of Kerman province, Iran. Data were collected with questioner method from expert farmers. Results indicated that the average of total input energy is about 84309.43 MJ ha−1 and the average of total output energy is 130217.14 MJ ha−1. Irrigation water (36%) and fertilizers (26%) were found to be the most important energy inputs in potato production. Unlike most literature reviews, in this study for better and more accurate model evolutions in energy forecasting, five different sizes of training selection (TS) were used: 50%, 60%, 70%, 80% and 90%. Some statistical indexes (RMSE, MAPE, and R2) of the different data selection calculated from k-fold in two training sets. The results showed that RBF model has a great prediction performance at all different values of training data selection. The average value of R2 was found to be more than 0.98. Between SVM and MLP models, the test performance will be improved by size reduction of training selection. Thus, the RBF model is chosen as the best model for fitting and modeling the output energy of potato production.

In this paper, we compare computational intelligence methods to analyze HIV in order to investigate which network is best suited for HIV classification. The methods analyzed are autoencoder multi-layer perceptron (MLP), autoencoder radial basis functions (RBF), support vector machines (SVM) and neuro-fuzzy models (NFM). The autoencoder multi-layer perceptron yields the highest accuracy of 92% amongst all the models studied. The autoencoder radial basis function model has the shortest computational time but yields one of the lowest accuracies of 82%. The SVM model yields the worst accuracy of 80%, as well as the worst computational time of 203s. The NFM yields an accuracy of 86%, which is the second highest accuracy. The NFM, however, offers rules, which gives interpretation of the data. The area under the receiver operating characteristics curve for the MLP model is 0.86 compared to an area under the curve of 0.87 for the RBF model, and 0.82 for the neuro- fuzzy model. The autoencoder MLP network model for HIV classification, is thus found to outperform the other network models and is a much better classifier.

Language Identification is an important issue in today's multilingual world. In this paper we have analyzed Fuzzy-SVM technique for identification of romanized plaintexts of five Indian regional languages namely Hindi, Bangla, Manipuri, Urdu and Kashmiri. Distinguishing features/characteristics have been extracted from romanized plaintexts of each of these five languages and represented suitably through Fuzzy Sets on a normalized scale. These normalized feature vectors are given as input to the Support Vector Machine (SVM) based classifier. For constructing the hyperplane in a higher dimension space Guassian Radial Basis Kernal function has been used. The proposed Pattern Recognition (PR) system is independent of the dictionaries of these languages and can even identify plaintext with unknown word boundaries. This PR system (Language Identifier) can be used for automatic segregation of plain texts of these languages while analyzing intercepted, multiplexed and interleaved Speech/Data/Fax communication. The proposed method significantly improves the classification accuracy compared to the other methods even for smaller text length messages.

Arabic language can be used by native and non- native speakers; due to Arabic is the language of the holy book of Muslims. In this paper, Arabic phoneme recognition system is proposed based on Malay speakers. This system consists of three main stages. The first stage is noise reduction and it aims to enhance the phoneme signals by excluding the unvoiced signals and keep only the voiced signal. Wiener filter is adapted to accomplish this task. The second stage is based on Mel-Frequency Cepstral Coefficients method to extract a vector of features to represent each phoneme signal. Eventually, pattern recognition neural network is designed as recognizer. The proposed system produces sufficient outcomes with 20 hidden neurons. Keywords—Arabic; Malay; pattern recognition; Wiener; Mel- Frequency Cepstral Coefficients I. INTRODUCTION It is well known the wide range of Artificial Neural Networks (ANN) application in speech recognition, financial, telecommunications, electronics and medical. As for speech recognition system for Arabic language, (1) & (2) have addressed the implementation of recognition system based on Arabic native speakers. In general, neural networks (NNs) for phonemes recognition are divided into several categories for instance Very Large Scale Integration (VLSI) NN that can be divided into digital and analogue with digital NNs being more compatible with feed-forward neural networks while analogue NNs are found to be more successful with recurrent NNs (3). From Automatic Speech System (ASR) point of view, NNs for recognition purpose can be divided into two main category namely conventional neural networks such as Multilayer Perceptron (MLP) and Radial Basis Function (RBF) and secondly Recurrent Neural Networks (RNN). The first category of NNs was implemented as pattern classifiers and proven capable for recognition purpose but not at par as compared to Hidden Markov Model (HMM) performance (2). At present, HMM has been proven to be the most successful approach in ASR research area until recently the possibility to hybrid both HMM and ANN (4) & (5). In general, speech recognition systems consist of three stages specifically noise reduction, feature extraction and classification. The aim of noise reduction is to reduce the level of noise in the speech signal and make it more compatible for the ASR. As for feature extraction, this process attempted to extract the most relevant features related to the phoneme signal. The extracted features acted as input features of the NN during training and testing. Finally, the purpose of classification process is to arrange the most similar and related features in categories based on the training phase. Basically, networks training database are being observed and updated during training session of the networks, involving the weights and biases arguments. Networks with lower errors can be considered as better networks (6). Hence, this paper deems to explore further the ability of NN for Arabic phonemes recognition spoken by Malay individuals. This paper is arranged as following; related work is discussed in section II followed by section III that discussed method to be implemented based on Wiener filter, Mel-Frequency Cepstral Coefficients and pattern recognition neural network. Section IV elaborated experimental analysis and results and followed by Conclusion section.

In protein fold recognition problem an effort is made to assign a fold to given proteins, this is of practical importance and has diverse application in the field of bioinformatics such as the discovery of new drugs, the individual implication of amino acid in a protein and bringing improvement in a specific protein function. In this paper, we have studied various machine learning techniques for protein fold recognition problem, and compared Support Vector Machine (SVM) with Radial Basis Function (RBF) kernel and Multilayer Perceptron (MLP) on a number of measures like the recognition accuracy of protein fold, the 10-fold cross validation accuracies and Kappa statistics. These techniques are applied to the well known Structural Classification of Proteins (SCOP) dataset in extensive experimentations. In this study Multilayer Perceptron (MLP) shows better accuracy on single protein feature (C, S, H, P, V, Z) of the SCOP dataset as compared to Support Vector Machine (SVM). A plausible reason of the better performance of MLP is that it uses all the available data for classification where as the SVM model cannot exploit all the available data.

This paper aims to investigate and compare the accuracy of different data mining classification schemes, employing Ensemble Machine Learning Techniques, for the prediction of heart disease. The Cleveland data set for heart diseases, containing 303 instances, has been used as the main database for the training and testing of the developed system. 10-Fold Cross-Validation has been applied in order to increase the amount of data, which would otherwise have been limited. Different classifiers, namely Decision Tree (DT), Naive Bayes (NB), Multilayer Perceptron (MLP), K-Nearest Neighbor (K-NN), Single Conjunctive Rule Learner (SCRL), Radial Basis Function (RBF) and Support Vector Machine (SVM), have been employed. Moreover, the ensemble prediction of classifiers, bagging, boosting and stacking, has been applied to the dataset. The results of the experiments indicate that the SVM method using the boosting technique outperforms the other aforementioned methods.

Research into the problem of classification of radar returns from the ionosphere has been taken up as a challenging task for the neural networks (NNs). It appears from the literature review that for the Multi layer Perceptron (MLP) NN trained with backpropagation, reported average classification accuracy was about 96% on the test instances. This paper investigates and designs an optimal classifier using a radial basis function (RBF) NN. Authors compare the performance of two NN configurations, namely a well-known MLP NN model and the proposed RBF NN model on the radar dataset collected from the published studies. It is shown that the proposed RBF NN, consistently, has 100% accuracy on bad instances and 99.1935% accuracy on good instances. The results show that the proposed RBF NN classifier clearly outperforms the MLP NN one in various performance measures such as MSE, NMSE, correlation coefficients, area under the ROC curve and classification accuracy on the testing datasets even after attempting different data partitions.

A real-time hand gesture classification model based on a robust feature extraction and machine learning algorithm is proposed in this paper. The Delsys Trigno Wireless EMG System is used to collect the EMG database. To improve the performance of EMG pattern recognition system, a new feature set based on bispectrum feature extraction is used as it can detect the non-Gaussian and nonlinear characteristics of EMG, unlike conventional feature set. The feature extracted is then fed to the classifiers to classify two hand gestures, hand close and hand open. The performance of the proposed pattern recognition system is measured in terms of their classification accuracies. The classifier algorithms used in this work are set of base classifiers such as Naive Bayes, support vector machine (SVM) and decision tree. Further to achieve increased classification accuracy ensemble classifiers such as random forest, gradient boost and AdaBoost are employed. From the experimental results, it is observed that the Adaboost performs better with high classification accuracy of 99% as compared to other classifiers such as SVM, decision tree, Naive Bayes, random forest and gradient boost with accuracies of 81%, 75%, 81%, 75% and 75%, respectively. Due to its adaptiveness in correcting the weak learners (base classifiers) in favor of reducing misclassification by the previous classifiers, AdaBoost resulted in higher accuracy compared to other classifiers.

In every country in the world, there are a number of amputees who have been exposed to some accidents that led to the loss of their upper limbs. The aim of this study is to suggest a system for real-time classification of five classes of shoulder girdle motions for high-level upper limb amputees using a pattern recognition system. In the suggested system, the wavelet transform was utilized for feature extraction, and the extreme learning machine was used as a classifier. The system was tested on four intact-limbed subjects and one amputee, with eight channels involving five electromyography channels and three-axis accelerometer sensor. The study shows that the suggested pattern recognition system has the ability to classify the shoulder girdle motions for high-level upper limb motions with 88.4% average classification accuracy for four intact-limbed subjects and 92.8% classification accuracy for one amputee by combining electromyography and accelerometer channels. The outcomes of this study may suggest that the proposed pattern recognition system can help to provide control signals to drive a prosthetic arm for high-level upper limb amputees.

Introduction: The prevalence of COVID-19 has significantly impacted work environments and the workforce. Therefore, identifying the most important preventive and control strategies, as well as assessing their effectiveness, is of paramount importance. Various studies have shown that machine learning algorithms can be used to predict complex and nonlinear issues, including predicting the behavior of various diseases such as COVID-19 and the parameters affecting it, and can be beneficial. The purpose of this study has been to examine the importance of preventive measures and hygiene behaviors in preventing COVID-19 in the oil refining industry using various machine learning models. Material and Methods: For this purpose, demographic information and health behaviors of individuals were collected. Subsequently, a multi-layer perceptron (MLP), radial basis function (RBF), and support vector machine (SVM) models were compared to enhance the analysis of the effects of preventive measures on COVID-19 infection. Finally, the most influential factors affecting the likelihood of COVID-19 infection were determined using sensitivity analysis. Results: The results showed that the accuracies achieved in predicting the impact of preventive measures and health behaviors on COVID-19 in occupational settings were 78.1%, 81.2%, and 78.1% by MLP, RBF, and SVM respectively. The RBF model was identified as the most accurate model for predicting the impact of health behaviors on COVID-19 disease Additionally, the level of social distancing with customers, handwashing frequency and disinfection, the availability of cleansing and disinfecting agents for hands and surfaces in the workplace, and gatherings for eating meals and snacks were identified as the most significant health behaviors influencing the prevalence of COVID-19 in the workplace. Conclusion: Studies of this nature can underscore the importance of attention to preventive measures and health behaviors in unprecedented circumstances. Furthermore, the utilization of artificial intelligence models and tools such as DSS (Decision Support Systems) can serve as powerful tools for optimizing control measures in work environments.