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
T
Tanh function,
1
,
315
-
316
back-propagation and,
54
error surface and,
115
,
118
,
119
,
123
,
128
generalization and,
251
for single-layer networks,
15
Target, effective,
278
-
281
,
289
Target outputs, introduction to,
7
-
10
Teacher,
7
,
11
,
12
-
14
,
49
Tessellation, Voronoi,
215
-
216
Threshold term,
17
-
18
.
See also
Linear threshold units
Tiling algorithm,
206
-
209
Trace(
H
),
81
-
82
Training.
See also
Algorithms
;
Supervised learning
definition of,
7
jitter and,
277
-
292
optimal amount of,
250
Training
(cont.)
as optimization problem,
155
over-,
198
,
237
,
241
-
242
,
249
-
251
,
292
steps in,
49
Training set size.
See
Size
Training time,
67
-
85
factors that increase,
68
-
69
initialization and,
97
,
105
,
106
momentum and,
74
-
77
,
85
-
95
pruning and,
219
,
237
scaling of,
68
-
70
training set size and,
68
,
71
,
199
variations to improve,
135
-
153
Two-hidden-layers networks,
32
-
33
,
38
-
39
,
109
Two-spirals problem,
108
,
144
,
254
Books24x7.com, Inc. © 1999-2001 –
Feedback