Presentation 2003/12/1
Bayesian Learning Adapted to State Transitions
Yoshifusa Ito, Cidambi Srinivasan, Hiroyuki Izumi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A Bayesian neural network can output approximate posterior probabilities of categories when its learning is completed. The outputs can be used as discriminant functions and an observable is allocated to the category with maximum posterior probability. We treat a neural network adapted to the source of observables which migrates among several states. A transition of state causes a change in the ratio of priors as well as that of posterior probabilities of categories. In this paper, we show that if an ordinary Bayesian neural network, having rather a small number of units, is supplied with a few additional input units, it can learn simultaneously state-wise distinct posterior probabilities. The learning algorithm and the result of simulations are also shown.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Bayesian decision / discriminant function / layered neural network / statetransition / approximation
Paper # NC2003-102
Date of Issue

Conference Information
Committee NC
Conference Date 2003/12/1(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Bayesian Learning Adapted to State Transitions
Sub Title (in English)
Keyword(1) Bayesian decision
Keyword(2) discriminant function
Keyword(3) layered neural network
Keyword(4) statetransition
Keyword(5) approximation
1st Author's Name Yoshifusa Ito
1st Author's Affiliation Department of Information and Policy Studies Aichi-Gakuin University()
2nd Author's Name Cidambi Srinivasan
2nd Author's Affiliation Department of Statistics University of Kentucky
3rd Author's Name Hiroyuki Izumi
3rd Author's Affiliation Department of Information and Policy Studies Aichi-Gakuin University
Date 2003/12/1
Paper # NC2003-102
Volume (vol) vol.103
Number (no) 490
Page pp.pp.-
#Pages 6
Date of Issue