Russell D. Reed and Robert J. Marks II NEURAL SMITHING   Mar 1999 ISBN 0-262-18190-8 352 pp. 105 illus. $57/£39.50 (hardback)   Preface 1.   Introduction 2.   Supervised Learning 3.   Single-Layer Networks 4.   MLP Representational Capabilities 5.   Back-Propagation 6.   Learning Rate and Momentum 7.   Weight-Initialization Techniques 8.   The Error Surface 9.   Faster Variations of Back-Propagation 10.   Classical Optimization Techniques 11.   Genetic Algorithms and Neural Networks 12.   Constructive Methods 13.   Pruning Algorithms 14.   Factors Influencing Generalization 15.   Generalization Prediction and Assessment 16.   Heuristics for Improving Generalization 17.   Effects of Training with Noisy Inputs   Appendices A   Linear Regression B   Principal Components Analysis C   Jitter Calculations D   Sigmoid-like Nonlinear Functions   References   Index Legal