Radial Basis Function Neural Network Classifier Research Articles

Classification of cancer types based on microRNA expression using a hybrid radial basis function and particle swarm optimization algorithm.

The diagnosis and treatment of cancer is one of the most challenging aspects of the medical profession, despite advances in disease diagnosis. MicroRNAs are small noncoding RNA molecules involved in regulating gene expression and are associated with several cancer types. Therefore, the analysis of microRNA data has become one of the most important areas of cancer research in recent years. This paper presents an improved method for cancer-type classification based on microRNA expression data using a hybrid radial basis function (RBF) and particle swarm optimization (PSO) algorithm. Two datasets containing microRNA information were used, and preprocessing and normalization operations were performed on the raw data. Feature selection was carried out by using the PSO algorithm, which can identify the most relevant and informative features in the data along with helping to prioritize them. Using a PSO algorithm for feature selection is an effective approach to microRNA analysis. This enhances the accuracy and reliability of cancer-type classifications based on microRNA expression data. In the proposed method, we, respectively, achieved an accuracy of 0.95% and 0.91% on both datasets, with an average of 0.93%, using an improved RBF neural network classifier. These results demonstrate that the proposed method outperforms previous works. RESEARCH HIGHLIGHTS: To enhance the accuracy of cancer-type classifications based on microRNA expression data. We present a minimal feature selection method using particle swarm optimization to reduce computational load & radial basis function to improve accuracy.

EEG signals classification using a new radial basis function neural network and jellyfish meta-heuristic algorithm.

The online version contains supplementary material available at 10.1007/s12065-022-00802-2.

Polysaccharide determination and habitat classification for fresh Dendrobiums with hyperspectral imagery and modified RBFNN.

This research aimed to study the visual and nondestructive detection of mannose (MN) and Dendrobium polysaccharides (DP) in Dendrobiums by using hyperspectral imaging technology. In order to determine the MN and DP concentrations nondestructively, we built radial basis function neural network (RBFNN) models based on NIR spectra (874–1734 nm) with a novel chemometric method to calculate the radial bases. And excellent results with the RP2 coefficients of 0.906 and 0.913 were obtained by the MN and DP detection models, respectively. In order to simplify the detection models based on full-range spectra, we designed an innovative genetic algorithm-successive projections algorithm (GA-SPA) strategy to extract the feature bands efficiently in two stages. Based on the feature bands selected by GA-SPA, we established the simplified detection models with the same high performance as those based on full-range spectra. By importing the feature bands of every pixel in the hyperspectral image into the simplified detection models, we successfully generated the distribution maps of MN and DP. Moreover, we also built an RBFNN classifier to categorize the habitats of Dendrobium. And the total classification accuracy reached 0.887. This research makes progress in Dendrobium quality evaluation and spectral detection technology.

Skin Lesion Classification Based on Surface Fractal Dimensions and Statistical Color Cluster Features Using an Ensemble of Machine Learning Techniques.

Simple SummaryThis study aimed to investigate the efficacy of implementation of novel skin surface fractal dimension features as an auxiliary diagnostic method for melanoma recognition. We therefore examined the skin lesion classification accuracy of the kNN-CV algorithm and of the proposed Radial basis function neural network model. We found an increased accuracy of classification when the fractal analysis is added to the classical color distribution analysis. Our results indicate that by using a reliable classifier, more opportunities exist to detect timely cancerous skin lesions.(1) Background: An approach for skin cancer recognition and classification by implementation of a novel combination of features and two classifiers, as an auxiliary diagnostic method, is proposed. (2) Methods: The predictions are made by k-nearest neighbor with a 5-fold cross validation algorithm and a neural network model to assist dermatologists in the diagnosis of cancerous skin lesions. As a main contribution, this work proposes a descriptor that combines skin surface fractal dimension and relevant color area features for skin lesion classification purposes. The surface fractal dimension is computed using a 2D generalization of Higuchi’s method. A clustering method allows for the selection of the relevant color distribution in skin lesion images by determining the average percentage of color areas within the nevi and melanoma lesion areas. In a classification stage, the Higuchi fractal dimensions (HFDs) and the color features are classified, separately, using a kNN-CV algorithm. In addition, these features are prototypes for a Radial basis function neural network (RBFNN) classifier. The efficiency of our algorithms was verified by utilizing images belonging to the 7-Point, Med-Node, and PH2 databases; (3) Results: Experimental results show that the accuracy of the proposed RBFNN model in skin cancer classification is 95.42% for 7-Point, 94.71% for Med-Node, and 94.88% for PH2, which are all significantly better than that of the kNN algorithm. (4) Conclusions: 2D Higuchi’s surface fractal features have not been previously used for skin lesion classification purpose. We used fractal features further correlated to color features to create a RBFNN classifier that provides high accuracies of classification.

An automatic Computer-Aided Diagnosis system based on the Multimodal fusion of Breast Cancer (MF-CAD)

The risk of death incurred by breast cancer is rising exponentially, especially among women. The early breast cancer diagnosis before it metastasizes helps medical staff controlling this disease, which decreases the risk of death. This made early breast cancer detection a crucial problem. Different imaging modalities offer complementary information concerning the same lesion helps to increase the performance of thcy fusing several modalities. This paper proposes a computerized features classification of breast cancer lesions through both the Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) and Digital Mammographic images (MGs). This study aims to investigate a Multimodal Fusion-based Computer-Aided Diagnosis (CAD) system, called MF-CAD, based on multivariate analysis of different modalities, for breast cancer mass detection. In this paper, firstly a new local feature descriptor is proposed in feature extraction, namely, the Gradient Local Information Pattern (GLIP), where we consider the gradient magnitude and orientation as well as the local differences as local binary features for DCE-MRI (or MGs) modality. Secondly, the fusion scheme is conducted using the Canonical Correlation Analysis (CCA) to highlight the intrinsic relation between these modalities. Finally, for comparative purposes, several selected machine learning classifiers (K-Nearest Neighbors, Support Vector Machine, Random forests, Artificial Neural Networks and Radial Basis Function Neural Network (RBFNN)) are used to distinguish between mass and No-mass breast images.Evaluation experiments of the diagnostic performances of our MF-CAD system are conducted over private datasets that contain both MG and DCE-MRI images acquired from 286 patients, which are “Breast DCE-MRI”, “Breast-MG” and “Breast Multimodal” datasets. Experimental results of the proposed MF-CAD system achieved an Area Under the ROC Curve (AUC) value of 99.10% using RBFNN classifier, while for each single modality alone, the best AUC values of 97.20% and 93.50% are obtained respectively for MG and DCE-MRI modalities using random forest classifier. A comparative study with recent existing approaches shows the competitive performances of the proposed approach.

Analysis and Classification of Faults in Switched Reluctance Motors Using Deep Learning Neural Networks

The simple and robust construction, less weight, wide operating speed range, and higher fault tolerance capability of switched reluctance (SR) motor make it a viable contender for the conventional dc and ac machines. The faults at the rotor, winding, stator converters, and sensors lead to overcurrent, increase torque ripples, and sudden breakdown of the system. Thus, it is an urgent requirement to recognize and classify the faults that exist in switched reluctance motors, thereby the reliability, robustness, and widespread utilization of SR motors can be increased. The phases of the SR Motor are excited by using an asymmetric bridgeless resonance converter. This paper proposes an automatic diagnosis and classification of faults using radial basis function neural network (RBFNN). A mathematical model of the SR motor is established to determine the state of the art of fault condition of SR motors. The speed data of SR motor are utilized by RBFNN to generate fault information. Gabor filter is used for preprocessing the input data, and segmentation is achieved using high accurate DCT-DOST transformation. A gray-level co-occurrence matrix optimized with a genetic algorithm is used to extract the features in the speed signal of the motor. A test setup was developed in MATLAB to measure the performances of the RBFNN classifier in real-time. The effectiveness of the simulated fault classification model is verified by comparing the results of the conventional PI controller with several optimized algorithm-based tuned PI controllers.

Discrimination of Focal and Non-Focal Seizures From EEG Signals Using Sliding Mode Singular Spectrum Analysis

Epilepsy is a neurological disorder, and it is diagnosed using electroencephalogram (EEG) signal. The discrimination of focal and non-focal categories of EEG signals is the primary task to locate epilepsy affected regions in the brain. In this paper, a novel approach to classify the focal and non-focal types of EEG signals is proposed. This approach is based on the decomposition of the EEG signal into reconstruction components (RCs) using sliding mode-singular spectrum analysis (SM-SSA). A total of five RCs are extracted from each EEG signal using SM-SSA. Then, a classifier is designed by combining the sparse-autoencoder (SAE) hidden layer, and radial basis function neural network (RBFN). Each RC obtained from the SM-SSA of EEG signal, and the SAE based RBFN (SAE-RBFN) classifier are used to classify the focal and non-focal types of EEG signals. The performance of the proposed approach is assessed using a publicly available database. The experimental results demonstrate that the third RC coupled with SAE-RBFN classifier produces an average accuracy, average sensitivity and average specificity values of 99.11%, 98.52%, and 99.70%, respectively using 10-fold cross-validation. The proposed approach is compared with existing methods for the discrimination of focal and non-focal EEG signals.

Soft computing-based method for estimation of almond kernel mass from its shell features

Fast and, at the same time, accurate methods for measuring physical properties of agricultural products are constantly needed to meet customer demand. Achieving evenly graded products based on their masses, as an important physical parameter, not only adds to customer satisfaction but also effects prices. The study analyzed the performance of the multi-linear regression (MLR) model and four different machine learning models—i.e. multilayer perceptron neural network (MLP-NN), radial basis function neural network (RBF-NN), adaptive neuro-fuzzy inference system (ANFIS) and support vector machine (SVM)—in order to compute the kernel mass of three almond varieties without breaking the shell. The best predicting model was used as the almond classifier into different mass classes (very small to large). The three main dimensions (length, maximum width, and maximum thickness) of nuts were manually measured and were used as the parameters for developing the models. Results showed that the intelligent system could successfully predict mass; however, RBF-NN (RMSE: 0.05-0.07 g; MAPE: 3.87–7.20 % for the training phase and RMSE: 0.05-0.06 g; MAPE: 3.84–6.77 % for the testing phase) had better results than other methods in prediction of the mass. The most overall accuracy of the RBF-NN classifier was 96.22 %. Considering the prediction accuracy and computational efficiency, RBF-NN is strongly recommended for non-destructive estimation of almond kernel mass.

ECG beat classification using empirical mode decomposition and mixture of features

Computer-aided analysis is useful in predicting arrhythmia conditions of the heart by analysing the recorded ECG signals. In this work, we proposed a method to detect, extract informative features to classify six types of heartbeat of ECG signals obtained from the MIT-BIH Arrhythmia database. The powerful discrete wavelet transform (DWT) is used to eliminate different sources of noises. Empirical mode decomposition (EMD) with adaptive thresholding has been used to detect precise R-peaks and QRS complex. The significant features consists of temporal, morphological and statistical were extracted from the processed ECG signals and combined to form a set of features. This feature set is classified with probabilistic neural network (PNN) and radial basis function neural network (RBF-NN) to recognise the arrhythmia beats. The process achieved better result with sensitivity of 99.96%, and positive predictivity of 99.81 with error rate of 0.23% in detecting the QRS complex. In class-oriented scheme, the arrhythmia conditions are classified with accuracy of 99.54%, 99.89% using PNN and RBF-NN classifier respectively. The obtained result confirms the superiority of the proposed scheme compared to other published results cited in literature.

Scene Classification Using Multi-Resolution WAHOLB Features and Neural Network Classifier

This article approaches scene classification problem by proposing an enhanced bag of features (BoF) model and a modified radial basis function neural network (RBFNN) classifier. The proposed BoF model integrates the image features extracted by histogram of oriented gradients, local binary pattern and wavelet coefficients. The extracted features are obtained in a hierarchical multi-resolution manner. The proposed approach is able to capture multi-level (the pixel-, patch-, and image-level) features. The histograms of features constructed by BoF model are then used for training a modified RBFNN classifier. As a modification, we propose using a new variant of particle swarm optimization, in which the parameters are updated adaptively, for determining the center of Gaussian functions in RBFNN. Experimental results demonstrate that our proposed approach significantly outperforms the state-of-the-art methods on scene classification of OT, FP, and LSP benchmark datasets.

Human Action Recognition Using Geometric Parameter Features and a Radial Basis Function Neural Network

A new biological method is proposed for the recognition of human actions based on geometric parameter features and the Radial Basis Function (RBF) neural network. First, the shape feature of the binary image sequences of a human body is calculated. Next, the geometric parameter features of the shape are computed based on the binary image sequences by using a vector which consists of the region and the shape feature. An improved RBF neural network classifier is then employed to identify the human action through posture recognition. To increase the learning efficiency of the neural network, improvements to the RBF learning algorithm are suggested to train the neural network. Experimental results indicate that the proposed scheme can effectively classify human actions and achieve a superior recognition rate.

Lung Cancer Classification Employing Proposed Real Coded Genetic Algorithm Based Radial Basis Function Neural Network Classifier.

A proposed real coded genetic algorithm based radial basis function neural network classifier is employed to perform effective classification of healthy and cancer affected lung images. Real Coded Genetic Algorithm (RCGA) is proposed to overcome the Hamming Cliff problem encountered with the Binary Coded Genetic Algorithm (BCGA). Radial Basis Function Neural Network (RBFNN) classifier is chosen as a classifier model because of its Gaussian Kernel function and its effective learning process to avoid local and global minima problem and enable faster convergence. This paper specifically focused on tuning the weights and bias of RBFNN classifier employing the proposed RCGA. The operators used in RCGA enable the algorithm flow to compute weights and bias value so that minimum Mean Square Error (MSE) is obtained. With both the lung healthy and cancer images from Lung Image Database Consortium (LIDC) database and Real time database, it is noted that the proposed RCGA based RBFNN classifier has performed effective classification of the healthy lung tissues and that of the cancer affected lung nodules. The classification accuracy computed using the proposed approach is noted to be higher in comparison with that of the classifiers proposed earlier in the literatures.

BeeRBF: A bee-inspired data clustering approach to design RBF neural network classifiers

Different methods have been used to train radial basis function (RBF) neural networks. This paper proposes the use of a bee-inspired algorithm, named cOptBees, plus a heuristic to automatically select the number, location and dispersions of basis functions to be used in RBF networks. cOptBees was originally designed to solve data clustering problems and the prototypes determined by the algorithm will be selected as the centers for the RBF network. The presented approach, named BeeRBF, is used to solve classification problems and is evaluated both in terms of the decision boundaries generated and classification accuracy. The performance of BeeRBF was compared with that of k-means, random center selection and some other proposals from the literature. The results show that BeeRBF is competitive and has the advantage of automatically determining the number of centers to be used in the RBF network.

R-Peak Detection using Daubechies Wavelet and ECG Signal Classification using Radial Basis Function Neural Network

This paper employed the Daubechies wavelet transform (WT) for R-peak detection and radial basis function neural network (RBFNN) to classify the electrocardiogram (ECG) signals. Five types of ECG beats: normal beat, paced beat, left bundle branch block (LBBB) beat, right bundle branch block (RBBB) beat and premature ventricular contraction (PVC) were classified. 500 QRS complexes were arbitrarily extracted from 26 records in Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database, which are available on Physionet website. Each and every QRS complex was represented by 21 points from p1 to p21 and these QRS complexes of each record were categorized according to types of beats. The system performance was computed using four types of parameter evaluation metrics: sensitivity, positive predictivity, specificity and classification error rate. The experimental result shows that the average values of sensitivity, positive predictivity, specificity and classification error rate are 99.8%, 99.60%, 99.90% and 0.12%, respectively with RBFNN classifier. The overall accuracy achieved for back propagation neural network (BPNN), multilayered perceptron (MLP), support vector machine (SVM) and RBFNN classifiers are 97.2%, 98.8%, 99% and 99.6%, respectively. The accuracy levels and processing time of RBFNN is higher than or comparable with BPNN, MLP and SVM classifiers.

Comparison of Intelligent Computing Techniques for Classification of Clinical EEG Signals

Objective: The objective of this work is to develop efficient classification systems using intelligent computing techniques for classification of normal and abnormal EEG signals. Methods: In this work, EEG recordings were carried out on volunteers (N=170). The features for classification of clinical EEG signals were extracted using wavelet transform and the feature selection was carried out using Principal Component Analysis. Intelligent techniques like Back Propagation Network (BPN), Adaptive Neuro-Fuzzy Inference System (ANFIS), Particle Swarm Optimization Neural network (PSONN) and Radial Basis function Neural network (RBFNN) were trained for diagnosing seizures. Further, the performance of the developed classifiers was compared. Results: Results demonstrate that RBFNN classifies normal and abnormal EEG signals better than the other methods. It appears that the RBFNN is able to detect Generalized Tonic-Clonic Seizure (GTCS) more efficiently than the Complex Partial Seizures (CPS). Positive predictive value was better in PSONN and ANFIS than BPN method. Conclusions: It appears that the combination of Wavelet transform method and PCA derived features along with RBFNN classifier is efficient for automated EEG signal classification

An Integrated Face Tracking and Facial Expression Recognition System

This article proposes a feature extraction method for an integrated face tracking and facial expression recognition in real time video. The method proposed by Viola and Jones [1] is used to detect the face region in the first frame of the video. A rectangular bounding box is fitted over for the face region and the detected face is tracked in the successive frames using the cascaded Support vector machine (SVM) and cascaded Radial basis function neural network (RBFNN). The haar-like features are extracted from the detected face region and they are used to create a cascaded SVM and RBFNN classifiers. Each stage of the SVM classifier and RBFNN classifier rejects the non-face regions and pass the face regions to the next stage in the cascade thereby efficiently tracking the face. The performance of tracking is evaluated using one hour video data. The performance of the cascaded SVM is compared with the cascaded RBFNN. The experiment results show that the proposed cascaded SVM classifier method gives better performance over the RBFNN and also the methods described in the literature using single SVM classifier [2]. While the face is being tracked, features are extracted from the mouth region for expression recognition. The features are modelled using a multi-class SVM. The SVM finds an optimal hyperplane to distinguish different facial expressions with an accuracy of 96.0%.

A Self-Adaptive RBF Neural Network Classifier for Transformer Fault Analysis

A new hybrid self-adaptive training approach-based radial basis function (RBF) neural network for power transformer fault diagnosis is presented in this paper. The proposed method is able to generate RBF neural network models based on fuzzy c-means (FCM) and quantum-inspired particle swarm optimization (QPSO), which can automatically configure network structure and obtain model parameters. With these methods, the number of neuron, centers and radii of hidden layer activated functions, as well as output connection weights can be automatically calculated. This learning method is proved to be effective by applying the RBF neural network in the classification of five benchmark testing data sets, and power transformer fault data set. The results clearly demonstrated the improved classification accuracy compared with other alternatives and showed that it can be used as a reliable tool for power transformer fault analysis.

An enhanced swarm intelligence clustering-based RBFNN classifier and its application in deep Web sources classification

The central problem in training a radial basis function neural network (RBFNN) is the selection of hidden layer neurons, which includes the selection of the center and width of those neurons. In this paper, we propose an enhanced swarm intelligence clustering (ESIC) method to select hidden layer neurons, and then, train a cosine RBFNN based on the gradient descent learning process. Also, we apply this new method for classification of deep Web sources. Experimental results show that the average Precision, Recall and F of our ESIC-based RBFNN classifier achieve higher performance than BP, Support Vector Machines (SVM) and OLS RBF for our deep Web sources classification problems.

An electronic nose and modular radial basis function network classifiers for recognizing multiple fragrant materials

An electronic nose with 16 TGS sensors is developed for effectively recognizing numerous kinds of odors. In order to solve large-sample, multi-class and high-dimensional classification problems, this paper proposes a type of modular radial basis function (RBF) network classifiers, in which every module consists of a single-layer RBF network and a single-layer perceptron. The method for optimally determining the number, locations and widths of RBF kernels and the target values of Gaussian activation functions is gone into details. The presented adaptive algorithm, which only propagates error one layer backwards, has much lower computational complexity than the back-propagation algorithm used in multilayer perceptrons. The electronic nose with the modular adaptive RBF neural network classifiers is able to reach the recognition rate of 96.67% for 21 kinds of simple and complex fragrant materials. The experimental result for the extend training set, which consists of 4050 samples and 84 classes, shows that the modular RBF networks as well as the adaptively learning algorithm have faster convergence rate, higher classification accuracy, larger probability to get optimal structures, and better ability to reach global minimum points, compared with the standard RBF networks and multilayer perceptrons. Therefore, the presented modular RBF networks are quite suitable for large sample problems.

Combining parametric and non-parametric algorithms for a partially unsupervised classification of multitemporal remote-sensing images

In this paper, we propose a classification system based on a multiple-classifier architecture, which is aimed at updating land-cover maps by using multisensor and/or multisource remote-sensing images. The proposed system is composed of an ensemble of classifiers that, once trained in a supervised way on a specific image of a given area, can be retrained in an unsupervised way to classify a new image of the considered site. In this context, two techniques are presented for the unsupervised updating of the parameters of a maximum-likelihood classifier and a radial basis function neural-network classifier, on the basis of the distribution of the new image to be classified. Experimental results carried out on a multitemporal and multisource remote-sensing data set confirm the effectiveness of the proposed system.