Presentation 2001/10/11
On the Approximation of Hyperbolic Dynamical Systems by Recurrent Neural Networks I.(Continuous Time Systems)
Ken-ichi Funabashi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Theoretical capability of continuous time recurrent neural networks is discussed in this paper. By the use of the approximation theorem on feedforward neural networks and the concept of pseudo-neural systems, mathematical base of approximation of dynamical systems by recurrent neural networks is constructed. Especially, approximation of dynamical systems with hyperbolic attractors by recurrent neural networks is discussed and some results of capability of recurrent neural networks are obtained in the consequence. These results give the theoretical base of application to identification of dynamical systems by recurrent neural networks.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) recurrent neural network / dynamical system / approximation / hyperbolic attractor / chaos
Paper # PRMU2001-104,NC2001-54
Date of Issue

Conference Information
Committee PRMU
Conference Date 2001/10/11(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 Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) On the Approximation of Hyperbolic Dynamical Systems by Recurrent Neural Networks I.(Continuous Time Systems)
Sub Title (in English)
Keyword(1) recurrent neural network
Keyword(2) dynamical system
Keyword(3) approximation
Keyword(4) hyperbolic attractor
Keyword(5) chaos
1st Author's Name Ken-ichi Funabashi
1st Author's Affiliation Center for Mathematical Sciences, The University of Aizu()
Date 2001/10/11
Paper # PRMU2001-104,NC2001-54
Volume (vol) vol.101
Number (no) 362
Page pp.pp.-
#Pages 7
Date of Issue