Skip to Book Content
Book cover image

Index

Neural Smithing: Supervised Learning in Feedforward Artificial Neural Networks
Russell D. Reed and Robert J. Marks II
Copyright © 1999 Massachusetts Institute of Technology
 

Index

A

Active layers, counting of, 31
Adaline networks, 29-30
Adaptive learning rate methods, 135-145, 147-148
benefits of, 72, 85, 95, 135-136, 151-152
limitations of, 152
Adaptive linear neurons. See Adaline networks
Adding units, in constructive methods, 198-199
Akaike's final prediction error (FPE), 260-261
Akaike's information criterion (AIC), 260-261
Algorithms. See also names of individual algorithms
back-propagation, 57-70
bias of, 249, 267
constructive (see Constructive methods)
factors in selection of, 156-158
generalization and, 157, 249-253
genetic, 178-179, 185-195
k-means, 107
LMS, 29, 90, 123, 124, 298
parameters for, 57, 62, 63, 71, 157, 185
perceptron learning, 23-27, 202, 205, 210-212
pocket, 28, 205
pruning (see Pruning methods)
robustness of, 157
Ancillary units, 208-209
AND function, 39, 109, 110
Approximation error, 37
Architecture selection, 197
Artificial neural networks
benefits of, 4-5
components of, 1
definition of, 1
ART networks, 198
Assumptions
error surface and, 120
optimization methods and, 179-180
Autoassociative networks, 12
Autoencoder networks, 12, 303-305
Axons, 1