Reconstructed dynamics and chaotic signal modeling

Author

Kuo, Jyh-Ming ; Principe, Jose C.

Author_Institution

Dept. of Electr. Eng., Florida Univ., Gainesville, FL, USA

Volume

5

fYear

1994

fDate

27 Jun-2 Jul 1994

Firstpage

3131

Abstract

A nonlinear AR model is derived from the reconstructed dynamics of a signal. The underlying system is assumed to be nonlinear, autonomous, and deterministic. In this formulation. The output error scheme is shown to be more suitable than the equation error scheme in network training. A method to incorporate the information of dynamical invariants in signal modeling is proposed. Using this global information, the authors are able to avoid the oscillation problem in training a network to model chaotic time series

Keywords

autoregressive processes; chaos; learning (artificial intelligence); neural nets; signal reconstruction; time series; chaotic signal modeling; dynamical invariants; equation error scheme; global information; network training; nonlinear AR model; output error scheme; reconstructed dynamics; Chaos; Delay effects; Frequency domain analysis; History; Multilayer perceptrons; Neural engineering; Nonlinear dynamical systems; Nonlinear equations; Predictive models; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on

Conference_Location

Orlando, FL

Print_ISBN

0-7803-1901-X

Type

conf

DOI

10.1109/ICNN.1994.374734

Filename

374734