Presentation 1996/2/3
An application of fluctuation-driven learning rule to asymmetrical cross-coupled neural networks
Masahito KAMIMURA, Takahumi OOHORI, Kazuhisa WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Fluctuation-driven learning (FDL) rule was originally intented for feedforward neural-networks (NN). In the NN for FDL, every neuron k contains an independent fluctuation. Going by the FDL rule, the weight from neuron j to k is modified by time average of product sequence of the output error from NN, the fluctuation in neuron k and the output from neuron j. The form of modification equation suggests the applicability of FDL rule to any type of NN. we show that under the FDL rule, the asymmetrical cross-coupled NN with self-connections can learn by means of the same modification equation as the feedforward NN. Furthermore, we clarify the effects of FDL, comparing with the learning rule of Boltzmann Machine.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Fluctuation-driven learning rule / Asymmetrical cross-coupled neural networks / Boltzmann machine / simulation
Paper # NC95-108
Date of Issue

Conference Information
Committee NC
Conference Date 1996/2/3(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) An application of fluctuation-driven learning rule to asymmetrical cross-coupled neural networks
Sub Title (in English)
Keyword(1) Fluctuation-driven learning rule
Keyword(2) Asymmetrical cross-coupled neural networks
Keyword(3) Boltzmann machine
Keyword(4) simulation
1st Author's Name Masahito KAMIMURA
1st Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology()
2nd Author's Name Takahumi OOHORI
2nd Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
3rd Author's Name Kazuhisa WATANABE
3rd Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
Date 1996/2/3
Paper # NC95-108
Volume (vol) vol.95
Number (no) 506
Page pp.pp.-
#Pages 8
Date of Issue