Hu Y.H., Hwang J.-N. (eds.) Handbook of Neural Network Signal Processing

CRC Press, 2002, -384 p.The field of artificial neural networks has made tremendous progress in the past 20 years in terms of theory, algorithms, and applications. Notably, the majority of real world neural network applications have involved the solution of difficult statistical signal processing problems. Compared to conventional signal processing algorithms that are mainly based on linear models, artificial neural networks offer an attractive alternative by providing nonlinear parametric models with universal approximation power, as well as adaptive training algorithms. The availability of such powerful modeling tools motivated numerous research efforts to explore new signal processing applications of artificial neural networks. During the course of the research, many neural network paradigms were proposed. Some of them are merely reincarnations of existing algorithms formulated in a neural network-like setting, while the others provide new perspectives toward solving nonlinear adaptive signal processing. More importantly, there are a number of emergent neural network paradigms that have found successful real world applications.
The purpose of this handbook is to survey recent progress in artificial neural network theory, algorithms (paradigms) with a special emphasis on signal processing applications. We invited a panel of internationally well known researchers who have worked on both theory and applications of neural networks for signal processing to write each chapter. There are a total of 12 chapters plus one introductory chapter in this handbook. The chapters are categorized into three groups. The first group contains in-depth surveys of recent progress in neural network computing paradigms. It contains five chapters, including the introduction, that deal with multilayer perceptrons, radial basis functions, kernel-based learning, and committee machines. The second part of this handbook surveys the neural network implementations of important signal processing problems. This part contains four chapters, dealing with a dynamic neural network for optimal signal processing, blind signal separation and blind deconvolution, a neural network for principal component analysis, and applications of neural networks to time series predictions. The third part of this handbook examines signal processing applications and systems that use neural network methods. This part contains chapters dealing with applications of artificial neural networks (ANNs) to speech processing, learning and adaptive characterization of visual content in image retrieval systems, applications of neural networks to biomedical image processing, and a hierarchical fuzzy neural network for pattern classification.Introduction to Neural Networks for Signal Processing
Signal Processing Using the Multilayer Perceptron
Radial Basis Functions
An Introduction to Kernel-Based Learning Algorithms
Committee Machines
Dynamic Neural Networks and Optimal Signal Processing
Blind Signal Separation and Blind Deconvolution
Neural Networks and Principal Component Analysis
Applications of Artificial Neural Networks to Time Series Prediction
Applications of Artificial Neural Networks (ANNs) to Speech Processing
Learning and Adaptive Characterization of Visual Contents in Image Retrieval Systems
Applications of Neural Networks to Image Processing
Hierarchical Fuzzy Neural Networks for Pattern Classification