Presentation 2001/3/10
Learning Process and Bifurcation in Recurrent Neural Networks
Y. Uwate, Y. Nakamura,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A change of phase status of a recurrent neural network occurs by applying of learning algorithm. In other words various kinds of bifurcation phenomena are observed in learning process. In this paper, we observed learning process for simple neural oscillator that bifurcation structure was clarified and relation between learning and bifurcations is considered. By using finite time interval learning a limit cycle is learned, and the learning path is illustrated in bifurcation diagram. As a result. we confirmed that vibration of learning happened at neighborhood of tangent bifurcation curves of equilibrium points and limit cycles. Moreover. in learning path along a tangent bifurcation curve of equilibrium point, it is confirmed that the case that did not arrive to a teacher point exists.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Recurrent neural network / Bifurcation phenomenon / Learning process / Limit cycle
Paper # NLP2000-168
Date of Issue

Conference Information
Committee NLP
Conference Date 2001/3/10(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) Learning Process and Bifurcation in Recurrent Neural Networks
Sub Title (in English)
Keyword(1) Recurrent neural network
Keyword(2) Bifurcation phenomenon
Keyword(3) Learning process
Keyword(4) Limit cycle
1st Author's Name Y. Uwate
1st Author's Affiliation Department of Electrical Engineering, Anan College of Technology()
2nd Author's Name Y. Nakamura
2nd Author's Affiliation Department of Electrical Engineering, Anan College of Technology
Date 2001/3/10
Paper # NLP2000-168
Volume (vol) vol.100
Number (no) 681
Page pp.pp.-
#Pages 6
Date of Issue