DocumentCode
296083
Title
On the identification of a chaotic system by means of recurrent neural state space models
Author
Suykens, J.A.K. ; Vandewalle, J.
Author_Institution
ESAT, Katholieke Univ., Leuven, Heverlee, Belgium
Volume
4
fYear
1995
fDate
Nov/Dec 1995
Firstpage
1570
Abstract
In this paper we discuss the identification of a chaotic system (double scroll attractor) by means of a simple recurrent neural state space model. A gradient based local optimization procedure is used, according to Narendra´s dynamic backpropagation. The orbit is learned by training in `packets´ of increasing time horizon and starting from a short time horizon
Keywords
backpropagation; chaos; identification; optimisation; recurrent neural nets; sensitivity analysis; state-space methods; Narendra´s dynamic backpropagation; chaotic system; double scroll attractor; gradient method; identification; local optimization; recurrent neural network; sensitivity model; state space models; time horizon; Backpropagation; Chaos; Cost function; Feedforward neural networks; Multi-layer neural network; Neural networks; Nonlinear systems; State-space methods; Terminology; Training data;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location
Perth, WA
Print_ISBN
0-7803-2768-3
Type
conf
DOI
10.1109/ICNN.1995.488851
Filename
488851
Link To Document