View Table of ContentsCreate a BookmarkAdd Book to My BookshelfPurchase This Book Online
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

Previous Section Next Section

Index

E

Early stopping, 265-266
Effective learning rate, 77, 87, 88, 90
Effective target, 278-281, 289
Eigenvalues, 127-128. See also Hessian matrix
initialization and, 106
learning rate and, 81-84, 85
node splitting and, 215
search then converge method and, 147
Energy
pruning and, 227-228
simulated annealing and, 177
Entropic error function, 123. See also Cross-entropy error function
Entropy, 271-272
Epochs, 58
Error function, 7, 37. See also E (t) curves
constructive methods and, 198-201, 205-206
cross-entropy, 9, 50, 110, 155, 276
entropic, 123
initialization and, 110
learning rate and, 71-80
LMS-threshold, 123, 124
penalty-term methods and, 220-221, 226-231
selection of, 253
sensitivity methods and, 220, 221-225
system energy and, 177
in training steps, 49-50, 52-69
Error surface
algorithm assumptions and, 120
characteristics of, 113-117
gain scaling and, 114, 132-134
Hessian matrix and, 127-132
learning rate and, 72
local minima of, 115-116, 117, 121-126
momentum and, 87-89
quadratic function and, 179
radial features of, 114-116
stair-steps on, 113-114, 121
total gradient of, 117
troughs and ridges of, 116-117
weight-space symmetries for, 118-119
E(t) curves
constructive methods and, 199, 201
learning rate and, 77-80
momentum and, 90-93
Evaluation-only methods
deterministic, 159-163
stochastic, 175-179
Evolutionary algorithm. See Genetic algorithm
Exclusive-OR function, 18, 19, 107, 251-253
Exploratory moves, 160

Top of current section
Previous Section Next Section
Books24x7.com, Inc. © 1999-2001  –  Feedback