Adaptive Pattern Recognition and Neural Networks

This book both introduces and relates the basic concepts of pattern recognition and neural networks. The first part provides a much-needed, current, and coherent view of pattern recognition. The author points to recent developments in several disciplines — artificial intelligence, cognitive science, computer engineering, neurobiology, philosophy, and psychology — and shows how these developments have advanced both the science and the technology of pattern recognition — and thereby advanced our understanding of human perception and cognition and our design of intelligent systems.

The second part of the book shows how adaptive pattern-recognition systems can be implemented using neural networks — that is, networks of elemental processors interconnected like their biological models. Neural-net implementations of pattern-recognition algorithms provide important, practical advantages by allowing iterative procedures to be implemented rapidly. But beyond those more immediate benefits, the author shows how a neural-net perspective on pattern information processing can stimulate us to be creative in visualizing new approaches.

In discussing algorithms, the book focuses on the generic issues involved and the insights provided to date— for example, how to deal with structures within patterns, with degrees of belief,, with the concept of time, and with the possibility of using adaptive pattern recognition for discovering simple solutions to some truly difficult problems. Students, researchers, and practitioners in engineering, robotics, and computer, behavioural, and biological sciences will all appreciate the broad and interdisciplinary perspective of this timely introduction to pattern recognition and neural-net computing.