Presentation 1997/2/6
Back-Propagation Learning of Infinite-Dimensional Dynamical Systems
Isao Tokuda, Kazuyuki Aihara,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper studies applicability of back-propagation learning techniques to recurrent neural networks with time-delays (DRNNs). The back-propagation learning is to teach spatib-temporal dynamics to the DRNNs. First, back-propagation learning algorithms are developed for DRNNs. The algorithms are then tested to teach periodic and chaotic dynamics to a class of DRNNs. Comparing with the back-propagation learning of ordinary recurrent neural networks having no time-delay (ORNNs), advantages and disadvantages of the back-propagation learning of DRNNs are discussed.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) back-propagation learning / recurrent neural network / time-delay / infinite-dimensional dynamical system
Paper # NLP96-138,NC96-92
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 ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Back-Propagation Learning of Infinite-Dimensional Dynamical Systems
Sub Title (in English)
Keyword(1) back-propagation learning
Keyword(2) recurrent neural network
Keyword(3) time-delay
Keyword(4) infinite-dimensional dynamical system
1st Author's Name Isao Tokuda
1st Author's Affiliation Department of Computer Science and Systems Engineering, Muroran Institute of Technology()
2nd Author's Name Kazuyuki Aihara
2nd Author's Affiliation Department of Mathematical Engineering and Information Physics, Faculty of Engineering, The University of Tokyo
Date 1997/2/6
Paper # NLP96-138,NC96-92
Volume (vol) vol.96
Number (no) 511
Page pp.pp.-
#Pages 6
Date of Issue