Издательство World Scientific, 1998. -316 pp.The field of artificial neural networks has come a long way. Depending on one’s perspective, one can trace its origins to the research on Mach bands in visual perception and its interpretation in terms of lateral inhibition, to Hodgkin and Huxley’s transmission line model of a nerve fiber, to McCulloch and Pitt’s model of a neuron, to Hebb’s postulate on learning, and to Rosenblatt’s Perceptron. A common theme in almost all of the earlier works is the desire to build a model of the brain. The scope of this research began to widen with the publication of McCulloch and Pitt’s model of a neuron. It is no accident that this model has an uncanny resemblance to the logic circuits of digital computers.
Although research in the computational capabilities of neural networks suffered a serious setback with the publication of Minsky and Pappert’s book on perceptrons, the almost simultaneous development of the back propagation algorithm in the 1970s by several investigators marked the turning point in the direction of neural network research. The publication of Hopfield’s paper, in 1983, in the Proceedings of the National Academy of Sciences (U.S.A.) almost caused a sensation in some circles. It is well over two decades now since the resurgence of neural network research. This is no longer a fringe area; it is mainstream.
This book is our way of celebrating the success of neural networks in different areas of engineering endeavor. The contributing authors come from many corners of the globe. All these chapters show how the power of neural networks can be exploited in modern engineering applications. Of the ten chapters, the first seven seem to form one group with an emphasis on image processing and industrial or manufacturing slant. Specifically, they touch on issues related to shape recognition, shape from shading, aircraft detection in SAR images, visualization of high-dimensional databases of industrial systems, 3-D object learning and recognition from multiple 2-D views, fingerprint classification and performance optimization in flexible manufacturing systems. The remaining three are applications to the communications area. Surprisingly, one of the earliest applications of neural networks was noise cancellation in telephone networks. That interest continues even today. The last three chapters included here address the issues involved in the exploding area of multimedia communications and in the area of mobile and cellular communications.On-Line Shape Recognition with Incremental Training Using a Neural Network with Binary Synaptic Weights
Neural Network Approaches to Shape from Shading
Neural Networks and Fuzzy Reasoning to Detect Aircraft in SAR Images
The Self-Organizing Map in Industry Analysis
A Self-Organizing Architecture for Invariant 3-D Object Learning and Recognition from Multiple 2-D Views
Industrial Applications of Hierarchical Neural Networks: Character Recognition and Fingerprint Classification
Neural Networks for Performance Optimization in Flexible Manufacturing Systems
Channel Assignment in Mobile Communication Networks – A Computational Intelligence Approach
Application of Cellular Compact Neural Networks in Digital Communication
Neural Networks for Process Scheduling in Communication Systems
