Zaknich A. Neural Networks for Intelligent Signal Processing

World Scientific, 2003, -510 p.The field of Artificial Neural Networks (ANN) represents an emerging design method still dominated by research. Despite the fact that there is much more research to be done before it becomes a fully accepted and established engineering discipline it is finding application in an ever- growing number of real-world problems. Enough fundamental principles have already been identified to form a good basis for practical scientific and engineering solutions. The main purpose of this book is to provide a sufficient understanding of ANNs to allow scientifically and mathematically informed students and professionals to begin applying them sensibly to practical problems encountered in both research and development activities. Another important purpose of this book is to introduce and provide an understanding of a very practical, and theoretically important, family of ANNs identified under the banner of Common Bandwidth Spherical Basis Function (CBSBF) ANNs.
The book can be used for introductory undergraduate or postgraduate engineering courses on "Artificial Neural Networks". However, it can also be used to provide useful short courses for science and mathematics students on selected topics. The content covers the fundamentals of ANNs from both theoretical and practical application perspectives. The specific aim has been to present a design approach, enough tools and sufficient understanding of significant ANN s to be able to use them to approach fundamental problems in Pattern Recognition (PR), Intelligent Signal Processing (ISP) and intelligent control. An adequate introductory course may exclude Chapters 9 and 10, and more advanced parts of Chapter
14. Such a course can be comfortably covered in one semester or approximately twenty six standard one hour lectures. A good understanding of calculus, vectors, matrix arithmetic and basic statistics is required to gain maximum benefit from such a course. It would also be helpful, but not strictly necessary, to have had some degree of exposure to fundamental pattern recognition, signal processing and control engineering ideas.A Brief Historical Overview
Basic Concepts
ANN Performance Evaluation
Basic Pattern Recognition Principles
ADALINES, Adaptive Filters, and Multi-Layer Perceptrons
Probabilistic Neural Network Classifier
General Regression Neural Network
The Modified Probabilistic Neural Network
Advanced MPNN Developments
Neural Networks Similar to the Common Bandwidth Spherical Basis Function Regression ANNs
Unsupervised Learning Neural Networks
Other Neural Network Models
Statistical Learning Theory
Application to Intelligent Signal Processing
Application to Intelligent Control
Discussion