DMCA
Four Types Of Learning Curves (1991)
| Venue: | Neural Computation |
| Citations: | 32 - 2 self |
BibTeX
@ARTICLE{Amari91fourtypes,
author = {Shun-ichi Amari and Shun-ichi Amari and Naotake Fujita and Naotake Fujita and Shigeru Shinomoto and Shigeru Shinomoto},
title = {Four Types Of Learning Curves},
journal = {Neural Computation},
year = {1991},
volume = {4},
pages = {608--618}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In learning from examples, the generalization error (t) is the average probability that an incorrect decision is made by a machine trained by t examples. The generalization error decreases as t increases, and the curve (t) is called a learning curve. The present paper uses the Bayesian approach to show that, under the random phase approximation (annealed approximation), learning curves are classified into four asymptotic types depending on situations. When a machine is deterministic with noiseless teacher signals, 1) ~ al when the correct machine parameter is unique in the parameter space, and 2) ~ at-2 when the set of the correct parameters has a finite measure. When teacher signals are noisy, 3) ~ at-l2 for a deterministic machine, and 4) ~ c + at 1 for a stochastic machine.
Keyphrases
learning curve generalization error parameter space annealed approximation teacher signal incorrect decision stochastic machine finite measure random phase approximation deterministic machine correct machine parameter bayesian approach average probability noiseless teacher signal asymptotic type present paper correct parameter
