Presentation 2003/1/28
Research of the Learning Process in a Recurrent Neural Network with Delayed Context Information
Satoshi SHIMADA, Kouji HARADA, Norio SHIRATORI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Nerve biology shows clearly that delay exists in the communication in a brain system, and some neural network models which took delay into consideration based on this fact are proposed. One of them, there are SRN (Simple Recurrent Network) proposed by Elman. SRN are multilayer perceptron networks. They can learn time series by outputs of the hidden layer fed back as inputs to that layer. The purpose of this research is clarifying the influence of delay to learning and generalization by using extended SRN model that can preserve delayed information from 1 to D steps. We used a^nb^n, one of the context-free language, related to natural language, as a learning sample. By this research, it became dear that introduction of delay is effective in learning and generalization.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) SRN / delay / context-free language / generalization
Paper # NLP2002-113
Date of Issue

Conference Information
Committee NLP
Conference Date 2003/1/28(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) Research of the Learning Process in a Recurrent Neural Network with Delayed Context Information
Sub Title (in English)
Keyword(1) SRN
Keyword(2) delay
Keyword(3) context-free language
Keyword(4) generalization
1st Author's Name Satoshi SHIMADA
1st Author's Affiliation Research Institute of Electrical Communication, Tohoku University()
2nd Author's Name Kouji HARADA
2nd Author's Affiliation Research Institute of Electrical Communication, Tohoku University
3rd Author's Name Norio SHIRATORI
3rd Author's Affiliation Research Institute of Electrical Communication, Tohoku University
Date 2003/1/28
Paper # NLP2002-113
Volume (vol) vol.102
Number (no) 626
Page pp.pp.-
#Pages 6
Date of Issue