Deep Learning Radial Basis Function Neural Networks Based Automatic Detection of Diabetic Retinopathy

Classification of Bisyllabic Lexical Stress Patterns Using Deep Neural Networks

This paper uses a radial basis function (RBF) neural network for star identification. A feature vector consisting of distance and angle cosine values is constructed as a library of recognition feature stars and input to the neural network. Then, the network’s output is the number of the corresponding primary star. Comparing the training efficiency and robustness of the grid algorithm with the BP neural network and radial basis function neural network for star identification, the experimental results show that the radial basis function neural network has a strong anti-interference ability against noise and pseudo-star interference. The following conclusions are drawn from the simulation tests: the sample recognition rate of radial basis function neural network algorithm can reach 96% for star position and star addition noise; when the proportion of pseudo-star reaches 20%, the sample recognition rate of radial basis function neural network algorithm is still higher than grid algorithm and BP neural network, which indicates that the radial basis function neural network algorithm has the solid anti-interference ability to both noise and pseudo-star. Therefore, the radial basis function neural network algorithm based on star Identification has specific application prospects.

In the process of moisture content's measurement with microwave resonant cavity based on RBF (radial basis function) neural network, the relationship among resonant frequency, quality factor and environmental temperature (the inputs of network) and moisture content (the output of network) is multiple non-linear. The performance of multiple non-linear regression algorithm is the major factor that determines the accuracy of measurement. A regression algorithm based on a RBF neural network is put forward to improve the measurement result in this paper. The RBF neural network regression algorithm can effectively avoid the BP algorithm disadvantages such as getting into local minimum and the low efficiency, and has the advantages such as high generalization precision and rapid convergence. Consequently, the measurement accuracy is enhanced. In this paper, experimental results show that the high-precision measurement value of moisture content can be obtained with the multiple non-linear regression algorithm based on RBF neural network.

Fault diagnosis for wireless sensor is very important to ensure signal acquisition precision. Radial basis function (RBF) neural network has strong classification ability. However, the selection of the connection weights, the hidden centers and the widths has an important influence on the classification performance of the RBF neural network in the learning process of RBF neural network. Thus, ant colony optimization is employ to gain the parameters of radial basis function neural network. Therefore, a novel method for fault diagnosis of wireless sensor based on RBF neural network and ant colony optimization is presented. The results of computational experiments show that ACO- RBF neural network has a great higher than RBF neural network.

Due to the advent of modern embedded systems and mobile devices with constrained resources, there is a great demand for incredibly efficient deep neural networks for machine learning purposes. There is also a growing concern of privacy and confidentiality of user data within the general public when their data is processed and stored in an external server which has further fueled the need for developing such efficient neural networks for real-time inference on local embedded systems. The scope of our work presented in this paper is limited to image classification using a convolutional neural network. A Convolutional Neural Network (CNN) is a class of Deep Neural Network (DNN) widely used in the analysis of visual images captured by an image sensor, designed to extract information and convert it into meaningful representations for real-time inference of the input data. In this paper, we propose a neoteric variant of deep convolutional neural network architecture to ameliorate the performance of existing CNN architectures for real-time inference on embedded systems. We show that this architecture, dubbed CondenseNeXt, is remarkably efficient in comparison to the baseline neural network architecture, CondenseNet, by reducing trainable parameters and FLOPs required to train the network whilst maintaining a balance between the trained model size of less than 3.0 MB and accuracy trade-off resulting in an unprecedented computational efficiency.

Diabetic retinopathy (DR) is a chronic eye disease for which early detection is highly essential to avoid any fatal results. Image processing of retinal images emerge as a feasible tool for this early diagnosis. Digital image processing techniques involve image classification which is a significant technique to detect the abnormality in the eye. Various automated classification systems have been developed in the recent years but most of them lack high classification accuracy. Artificial neural networks are the widely preferred artificial intelligence technique since it yields superior results in terms of classification accuracy. In this work, Radial Basis function (RBF) neural network based bi-level classification system is proposed to differentiate abnormal DR Images and normal retinal images. The results are analyzed in terms of classification accuracy, sensitivity and specificity. A comparative analysis is performed with the results of the probabilistic classifier namely Bayesian classifier to show the superior nature of neural classifier. Experimental results show promising results for the neural classifier in terms of the performance measures.

Digital image recognition of plant diseases could reduce the dependence of agricultural production on the professional and technical personnel in plant protection field and is conducive to the development of plant protection informatization. In order to find out a method to realize image recognition of plant diseases, four kinds of neural networks including backpropagation (BP) networks, radial basis function (RBF) neural networks, generalized regression networks (GRNNs) and probabilistic neural networks (PNNs) were used to distinguish wheat stripe rust from wheat leaf rust and to distinguish grape downy mildew from grape powdery mildew based on color features, shape features and texture features extracted from the disease images. The results showed that identification and diagnosis of the plant diseases could be effectively achieved using BP networks, RBF neural networks, GRNNs and PNNs based on image processing. For the two kinds of wheat diseases, the best prediction accuracy was 100% with the fitting accuracy equal to 100% while BP networks, GRNNs or PNNs were used, and the best prediction accuracy was 97.50% with the fitting accuracy equal to 100% while RBF neural networks were used. For the two kinds of grape diseases, the best prediction accuracy was 100% with the fitting accuracy equal to 100% while BP networks, GRNNs or PNNs were used, and the best prediction accuracy was 94.29% with the fitting accuracy equal to 100% while RBF neural networks were used.

Automated image classification is an essential task of the computer vision field. The tagging of images into a set of predefined groups is referred to as image classification. The implementation of computer vision to automate image classification would be beneficial because manual image evaluation and identification can be time-consuming, particularly when there are many images of different classes. Deep learning approaches are proven to overperform existing machine learning techniques in many fields in recent years, and computer vision is one of the most notable examples. The very deep neural network (VDCNN) is a powerful deep learning model for image classification, and this paper examines it briefly using MNIST handwritten digit dataset. This dataset is used to prove the efficacy of very deep neural networks over other deep learning models. The proposed study aims to comprehend the very deep neural network architecture used to accomplish a handwritten digit recognition task. The feasibility of the proposed model is evaluated using mean accuracy, validation accuracy, and standard deviation. The study results of the very deep neural network model are compared to a convolutional neural network and convolutional neural network with batch normalization. According to the results of the comparison study, very deep neural networks achieve high accuracy of 99.1% for a handwritten dataset. The outcome of the proposed work is used to interpret how well a very deep neural network performs when compared to the other two models of deep neural networks. This proposed architecture may be used to automate the classification of handwritten digits dataset.

Heartbeat classification is a crucial step for arrhythmia diagnosis during electrocardiographic (ECG) analysis. The new scenario of wireless body sensor network- (WBSN-) enabled ECG monitoring puts forward a higher-level demand for this traditional ECG analysis task. Previously reported methods mainly addressed this requirement with the applications of a shallow structured classifier and expert-designed features. In this study, modified frequency slice wavelet transform (MFSWT) was firstly employed to produce the time-frequency image for heartbeat signal. Then the deep learning (DL) method was performed for the heartbeat classification. Here, we proposed a novel model incorporating automatic feature abstraction and a deep neural network (DNN) classifier. Features were automatically abstracted by the stacked denoising auto-encoder (SDA) from the transferred time-frequency image. DNN classifier was constructed by an encoder layer of SDA and a softmax layer. In addition, a deterministic patient-specific heartbeat classifier was achieved by fine-tuning on heartbeat samples, which included a small subset of individual samples. The performance of the proposed model was evaluated on the MIT-BIH arrhythmia database. Results showed that an overall accuracy of 97.5% was achieved using the proposed model, confirming that the proposed DNN model is a powerful tool for heartbeat pattern recognition.

Chapter 2 - Intelligent Data Analytics for PV Fault Diagnosis Using Deep Convolutional Neural Network (ConvNet/CNN)

Breast cancer is a fatal disease that affects millions of women worldwide. The number of cases of breast cancer has risen over time. Although it is difficult to prevent this disease, the survival rate can be improved with early detection and proper treatment planning of the disease. Thanks to breakthroughs in deep learning, the progress of computer-assisted diagnosis (CAD) of breast cancer has seen a lot of improvement. Deep neural networks have advanced to the point where their diagnostic capabilities are approaching those of a human specialist. Mammograms are crucial radiological images used to detect breast cancer at its early stage. Mammogram image scaling at the input layer is required for training deep convolutional neural networks (DCNNs) directly on high-resolution images. This could result in the loss of crucial information for discovering medical abnormalities. Instead of developing an image classifier, the idea is to create a patch classifier. The technique for extracting abnormal patches from mammography images is proposed in this research. Patches are extracted from a benchmark and publicly available MIAS dataset. These patches are then used to train deep learning classifiers such as VGG-16, ResNet-50, and EfficientNet-B7. With the patches already included in the CBIS-DDSM dataset, we contrast the results of the MIAS patches. EfficientNet-B7 on CBIS-DDSM patches produced good results (92% accuracy) when compared to other classifiers such as VGG-16 and ResNet-50. We also discovered that ResNet-50 is demonstrated to be quite robust on both datasets.

Classical radial basis function (RBF) neural network directly projects input samples into a high dimension feature space through some radial basis functions, and does not take account of the high-order statistical relationship among variables of input samples. But the high-order statistical relationship does play an important part in pattern recognition (classification) area. In order to take advantage of the high-order statistical relationship among variables of input samples in neural network, a novel independent radial basis function (IRBF) neural network is proposed in this paper. Then a new hybrid system combining multiresolution analysis, principal component analysis (PCA) and our proposed IRBF neural network is also proposed for face recognition. According to experiments on FERET face database, our proposed approach could outperform newly proposed ICA algorithm. And it is also confirmed that our proposed approach is more robust to facial expression, illumination and aging than ICA in face recognition.

Video hyperlinking represents a classical example of multimodal problems. Common approaches to such problems are early fusion of the initial modalities and crossmodal translation from one modality to the other. Recently, deep neural networks, especially deep autoencoders, have proven promising both for crossmodal translation and for early fusion via multimodal embedding. A particular architecture, bidirectional symmetrical deep neural networks, have been proven to yield improved multimodal embeddings over classical autoencoders, while also being able to perform crossmodal translation. In this work, we focus firstly at evaluating good single-modal continuous representations both for textual and for visual information. Word2Vec and paragraph vectors are evaluated for representing collections of words, such as parts of automatic transcripts and multiple visual concepts, while different deep convolutional neural networks are evaluated for directly embedding visual information, avoiding the creation of visual concepts. Secondly, we evaluate methods for multimodal fusion and crossmodal translation, with different single-modal pairs, in the task of video hyperlinking. Bidirectional (symmetrical) deep neural networks were shown to successfully tackle downsides of multimodal autoencoders and yield a superior multimodal representation. In this work, we extensively tests them in different settings, with different single-modal representations, within the context of video-hyperlinking. Our novel bidirectional symmetrical deep neural networks are compared to classical autoencoders and are shown to yield significantly improved multimodal embeddings that significantly (alpha=0.0001) outperform multimodal embeddings obtained by deep autoencoders with an absolute improvement in precision at 10 of 14.1% when embedding visual concepts and automatic transcripts and an absolute improvement of 4.3% when embedding automatic transcripts with features obtained with very deep convolutional neural networks, yielding 80% of precision at 10.

To improve the reservoir long-term runoff forecasting accuracy. Adaptive regulation ant colony system algorithm (ARACS) is proposed. The forecast model is set up by using an adaptive regulation ant colony system algorithm and the radial basis function (RBF) neural network combined to form ARACS-RBF hybrid algorithm. Form the reservoir long-term runoff forecast model based on the hybrid algorithm. Then carry out the reservoir long-term runoff forecast by using the method and history runoff data. The result shows the convergence of method is faster and forecast accuracy is more accurate than that of the traditional ant colony system algorithm-RBF neural network and RBF neural network. The method improves forecast accuracy and improves the RBF neural network generalization capacity; it has a high computational precision, and in 98% of confidence level the average percentage error is not more than 6%. The hybrid algorithm can be used efficaciously in long-term runoff forecasting of the reservoir and river.

In this paper, we are going to propose an online radial basis function (RBF) neural network algorithm without any preprocessing step. Then a kernel principal component analysis (KPCA) is coupled with the proposed online RBF neural network algorithm. Indeed, the KPCA method is used as a preprocessing step to reduce the feature dimension which fed to the RBF neural network. Reducing memory requirements of the models makes RBF neural network training efficient and fast. These two proposed algorithms are applied, with success, for identification of a mobile robot position. The simulation results present that the used sigmoid function as a kernel, compared to other kernel functions, which gives an excellent model and a minimum mean square error.