Presentation 1998/10/24
Memorization of a pattern-dynamics by learning a symbol-dynamics
Fumihiko Ishida, Yoshinori Hayakawa, Yoshihiro Hayakawa, Yasuji Sawada,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a neural network architecture that memorizes a pattern (external information)-dynamics by a learning symbol (internal representation)-dynamics to realize 'prediction' of the external world, which is one of the important brain function. The architecture has two feedforward neural networks to encode patterns or to decode symbols and a recurrent neural network to memorize a symbol-dynamics. In this paper, we present a new algorithm to memorize time-series to realize the symbol-dynamics. Computer simulation shows that it is possible to memorize a rotating bar, as an example of a pattern-dynamics.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) pattern-dynamics / symbol-dynamics / encoding / decoding / time-series memorization / neural networks
Paper # NC98-47
Date of Issue

Conference Information
Committee NC
Conference Date 1998/10/24(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Memorization of a pattern-dynamics by learning a symbol-dynamics
Sub Title (in English)
Keyword(1) pattern-dynamics
Keyword(2) symbol-dynamics
Keyword(3) encoding
Keyword(4) decoding
Keyword(5) time-series memorization
Keyword(6) neural networks
1st Author's Name Fumihiko Ishida
1st Author's Affiliation Graduate School of information Science, Tohoku Univ.()
2nd Author's Name Yoshinori Hayakawa
2nd Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.
3rd Author's Name Yoshihiro Hayakawa
3rd Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.
4th Author's Name Yasuji Sawada
4th Author's Affiliation Research Institute of Electrical Communication, Tohoku Univ.
Date 1998/10/24
Paper # NC98-47
Volume (vol) vol.98
Number (no) 365
Page pp.pp.-
#Pages 7
Date of Issue