Learning Process of a Stochastic Feed-Forward Neural Network

Sumiyoshi Fujiki; Nahomi M. Fujiki

doi:10.1143/jpsj.64.757

Journal of the Physical Society of Japan

Online ISSN : 1347-4073
Print ISSN : 0031-9015
ISSN-L : 0031-9015

Learning Process of a Stochastic Feed-Forward Neural Network

Sumiyoshi Fujiki, Nahomi M. Fujiki

著者情報

キーワード: a learning process, feed-forward neural network, stochastic layered network

ジャーナル認証あり

1995 年 64 巻 3 号 p. 757-765

DOI https://doi.org/10.1143/jpsj.64.757

詳細

抄録

A positive reinforcement type learning algorithm is formulated for a stochastic feed-forward neural network by minimizing a relative entropic measure, and a learning equation similar to that of the Boltzmann machine is obtained. The learning of the network actually shows a similar result to that of the Boltzmann machine in the classification problems of AND and XOR, by numerical experiments.