Page 326 -

316 Index

edit hard limiter, 158
costs, 266 Hebb rule, 170, 228
operations, 266, 281 Hessian matrix, 186
edition method, 1 15 hidden neuron, 167
eigenvalues, 38 hierarchical
eigenvectors, 38, 65, 121 classifier, 130
elitism, 209 network, 232
empirical error, 200 histogram, 42
empirical risk, 97 holdout method, 128
ensemble network, 233 homomorphism, 279
entropy, 137 Hopfield network, 169, 226, 275
epoch, 152, 175 Hough transform, 247
ERM, 97 hyperellisoid, 85
error hyperplane, 22, 80
bias, 127, 189 hy persphere, 30
energy, 148
mean, 184 icicle plot, 58
surface, 150 ID3 algorithm, 137
test set, 106 in-degree, 248
training set, 80, 106 interpolating kernel, 11 1
variance, 106, 127, 189 interpolation, 2 13
Euclidian classifier, 84 interquartile, 42
exhaustive search, 122 IPS, 175
external variables, 181 isodata, 70
isomorphism, 280
factor
analysis, 65 jogging weights, 178
loadings, 65
fat shattering, 198 Kaiser criterion, 40
feature kernel function, 213, 222
extraction, 16 k-means clustering, 70
selection, 121 k-NN, 1 14
space, 26 Kohonen network, 169,223
standardization, 56 Kolmogorov-Smirnov test, 43
vector, 13 Kruskal-Wallis test, 44
features, 12
FHR, 260 Lagrange multipliers, 2 17
finite state machine, 257 learning
Fisher discriminant, 88, 186 curve, 166
fitness factor, 208 rate, 15 1
forecasting, 181 size (bounds), 196
forward search, 123 least mean square, 149
leave-one-out method, 128
Gaussian distribution, 98 Levenberg-Marquardt, 205
genetic algorithm, 123, 208 Levenshtein distance, 266
gradient descent, 15 1 likelihood, 91
graph, 248 Lillefors correction, 43
edge, 248 line search, 202
node, 248 linear
growth function, 194 discriminant, 22, 83, 147
network, 148
Hamming distance, 228 transformation, 3 1