Presentation 1997/2/6
Experimental Study on Applicability of Fluctuation-Driven Learning Rule to Various Tasks
Hirotada FURUKAWA, Yoshinori UEMURA, Masahiro KIDOGUCHI, Takahumi OOHORI, Kazuhisa WATANABE,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Previously we proposed fluctuation-driven learning(FDL) rule for feedforward neural networks. According to the rule without error back-propagating, all connections are modified in parallel. Therefore, the rule is regarded as applicable widely to various tasks by various types of networks. We show experimentally that by changing the error function and learning parameters, FDL rule becomes applicable to the inverse modelling of systems with shifting function, the robot arm controlling by reinforcement learning, the flip-flop circuit tasks, the PSK modulating/demodulating, the swinging-up a pendulum by TD(λ) learning and the oscillating of sinusoidal waveform.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) fluctuation-driven learning rule / inverse model / reinforcement learning / TD(λ) / flip-flop circuit
Paper # NLP96-128,NC96-82
Date of Issue

Conference Information
Committee NC
Conference Date 1997/2/6(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Experimental Study on Applicability of Fluctuation-Driven Learning Rule to Various Tasks
Sub Title (in English)
Keyword(1) fluctuation-driven learning rule
Keyword(2) inverse model
Keyword(3) reinforcement learning
Keyword(4) TD(λ)
Keyword(5) flip-flop circuit
1st Author's Name Hirotada FURUKAWA
1st Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology()
2nd Author's Name Yoshinori UEMURA
2nd Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
3rd Author's Name Masahiro KIDOGUCHI
3rd Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
4th Author's Name Takahumi OOHORI
4th Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
5th Author's Name Kazuhisa WATANABE
5th Author's Affiliation Department of Electrical Engineering, Hokkaido Institute of Technology
Date 1997/2/6
Paper # NLP96-128,NC96-82
Volume (vol) vol.96
Number (no) 511
Page pp.pp.-
#Pages 8
Date of Issue