Presentation 1996/3/19
A Modified back-Propagation Algorithm that Automatically Removes Redundant Hidden Units by Competition
Yutaka MATSUNAGA, Kazuyuki MURASE, Osamu yamakawa, Manabu TANIFUJI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We present an optimization algorithm for layered neural networks that can determine both the connection weights and the number of hidden units at the same time during learning process. It uses a mechanism of competition among hidden units that is based upon "the goodness factor (GF)". The GF value, calculated for each hidden unit, represents the importance of the hidden unit for the upper layer units to separate their input space. In all the learning problems tested, the minimal or nearly minimal networks for givenproblems were obtained with high probability, and the learning time is faster than the conventional back-propagation algorithm.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) layered neural network / back propagation / competition / optimization / association units
Paper # NC95-170
Date of Issue

Conference Information
Committee NC
Conference Date 1996/3/19(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) A Modified back-Propagation Algorithm that Automatically Removes Redundant Hidden Units by Competition
Sub Title (in English)
Keyword(1) layered neural network
Keyword(2) back propagation
Keyword(3) competition
Keyword(4) optimization
Keyword(5) association units
1st Author's Name Yutaka MATSUNAGA
1st Author's Affiliation Department of Information Science Faculty of Engineering, Fukui University()
2nd Author's Name Kazuyuki MURASE
2nd Author's Affiliation Department of Information Science Faculty of Engineering, Fukui University
3rd Author's Name Osamu yamakawa
3rd Author's Affiliation Center for Information Science, Fukui Prefectural University
4th Author's Name Manabu TANIFUJI
4th Author's Affiliation Department of Information Science Faculty of Engineering, Fukui University
Date 1996/3/19
Paper # NC95-170
Volume (vol) vol.95
Number (no) 599
Page pp.pp.-
#Pages 8
Date of Issue