Presentation 2006-12-13
A comparison between correlation learning and pseudo-relaxation learning for a complex-valued bidirectional associative model
Ryuuichi SASAKI, Masaharu ADACHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Bidirectional associative memories have been proposed by Kosko. Kosko's model consists of two layers of neurons with synaptic connections between layers. In the original model, the synaptic weight connections are determined by a correlation matrix of the stored patterns. However, when the training pattern vectors are not orthogonal, cross-talk noise may occur in the system. To solve such problem, Oh.et.al. have been proposed Pseudo-relaxation learning algorithm for BAM (PRLAB) which uses projection method. PRLAB dose not require orthgonality or any special encoding of the training pairs and extremely increase the storage capacity. We extend the real number PRLAB algorithm to complex number. In the case that the complex-valued BAM trained by the extended PRLAB, we show that the storage capacity is increased as the same as in the real number PRLAB. Moreover, it is also shown that when the complex-valued BAM fails to store the patterns to have high storage capacity, the time series of directional cosine during the training shows non-periodic oscillation. An estimation of Lyapunov exponents of the time series suggests that such an oscillation may be deterministic chaos.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) bidirectional associative memory / complex-valued neural network / correlation learning / Pseudo-relaxation learning algorithm for BAM (PRLAB)
Paper # NLP2006-93
Date of Issue

Conference Information
Committee NLP
Conference Date 2006/12/6(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Nonlinear Problems (NLP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A comparison between correlation learning and pseudo-relaxation learning for a complex-valued bidirectional associative model
Sub Title (in English)
Keyword(1) bidirectional associative memory
Keyword(2) complex-valued neural network
Keyword(3) correlation learning
Keyword(4) Pseudo-relaxation learning algorithm for BAM (PRLAB)
1st Author's Name Ryuuichi SASAKI
1st Author's Affiliation Department of Electronic Engineering, Graduate School of Engineering, Tokyo Denki University()
2nd Author's Name Masaharu ADACHI
2nd Author's Affiliation Department of Electronic Engineering, Graduate School of Engineering, Tokyo Denki University
Date 2006-12-13
Paper # NLP2006-93
Volume (vol) vol.106
Number (no) 413
Page pp.pp.-
#Pages 6
Date of Issue