A neural network model for prediction error identification

N. Z. Hakim; J. J. Kaufman; R. S. Siffert; H. E. Meadows

doi:10.1109/acssc.1989.1200762

A neural network model for prediction error identification

N. Z. Hakim, J. J. Kaufman, R. S. Siffert, H. E. Meadows

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

A class of neural-network based nonlinear models suitable for application in the prediction error systems identification method is presented. The parameters of this class of models include a local interconnection neighborhood size, a time constant characterizes the neurons, a weight matrix, and input/output interconnection matrices. Some examples of the system behavior for different values of these parameters are presented to evaluate the suitability of the neural network prediction error model to fit a wide class of systems. The method is then applied to train the net to perform three classical problems: generation of Van der Pol oscillations, estimation of a sum of cosines in additive white Gaussian noise, and prediction of a chaotic time series generated by the logistic function.

Original language	English
Title of host publication	Conference Record - Asilomar Conference on Circuits, Systems & Computers
Publisher	Publ by Maple Press, Inc
Pages	119-122
Number of pages	4
ISBN (Print)	0929029301
DOIs	https://doi.org/10.1109/acssc.1989.1200762
State	Published - 1989
Externally published	Yes
Event	Twenty-Third Annual Asilomar Conference on Signals, Systems & Computers - Pacific Grove, CA, USA Duration: 30 Oct 1989 → 1 Nov 1989

Publication series

Name	Conference Record - Asilomar Conference on Circuits, Systems & Computers
Volume	1
ISSN (Print)	0736-5861

Conference

Conference	Twenty-Third Annual Asilomar Conference on Signals, Systems & Computers
City	Pacific Grove, CA, USA
Period	30/10/89 → 1/11/89

Access to Document

10.1109/acssc.1989.1200762

Cite this

Hakim, N. Z., Kaufman, J. J., Siffert, R. S., & Meadows, H. E. (1989). A neural network model for prediction error identification. In Conference Record - Asilomar Conference on Circuits, Systems & Computers (pp. 119-122). (Conference Record - Asilomar Conference on Circuits, Systems & Computers; Vol. 1). Publ by Maple Press, Inc. https://doi.org/10.1109/acssc.1989.1200762

@inproceedings{dba7ecc58ce045508182953e6829184d,

title = "A neural network model for prediction error identification",

abstract = "A class of neural-network based nonlinear models suitable for application in the prediction error systems identification method is presented. The parameters of this class of models include a local interconnection neighborhood size, a time constant characterizes the neurons, a weight matrix, and input/output interconnection matrices. Some examples of the system behavior for different values of these parameters are presented to evaluate the suitability of the neural network prediction error model to fit a wide class of systems. The method is then applied to train the net to perform three classical problems: generation of Van der Pol oscillations, estimation of a sum of cosines in additive white Gaussian noise, and prediction of a chaotic time series generated by the logistic function.",

author = "Hakim, {N. Z.} and Kaufman, {J. J.} and Siffert, {R. S.} and Meadows, {H. E.}",

year = "1989",

doi = "10.1109/acssc.1989.1200762",

language = "English",

isbn = "0929029301",

series = "Conference Record - Asilomar Conference on Circuits, Systems & Computers",

publisher = "Publ by Maple Press, Inc",

pages = "119--122",

booktitle = "Conference Record - Asilomar Conference on Circuits, Systems & Computers",

note = "Twenty-Third Annual Asilomar Conference on Signals, Systems &amp; Computers ; Conference date: 30-10-1989 Through 01-11-1989",

}

Hakim, NZ, Kaufman, JJ, Siffert, RS & Meadows, HE 1989, A neural network model for prediction error identification. in Conference Record - Asilomar Conference on Circuits, Systems & Computers. Conference Record - Asilomar Conference on Circuits, Systems & Computers, vol. 1, Publ by Maple Press, Inc, pp. 119-122, Twenty-Third Annual Asilomar Conference on Signals, Systems & Computers, Pacific Grove, CA, USA, 30/10/89. https://doi.org/10.1109/acssc.1989.1200762

A neural network model for prediction error identification. / Hakim, N. Z.; Kaufman, J. J.; Siffert, R. S. et al.
Conference Record - Asilomar Conference on Circuits, Systems & Computers. Publ by Maple Press, Inc, 1989. p. 119-122 (Conference Record - Asilomar Conference on Circuits, Systems & Computers; Vol. 1).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A neural network model for prediction error identification

AU - Hakim, N. Z.

AU - Kaufman, J. J.

AU - Siffert, R. S.

AU - Meadows, H. E.

PY - 1989

Y1 - 1989

N2 - A class of neural-network based nonlinear models suitable for application in the prediction error systems identification method is presented. The parameters of this class of models include a local interconnection neighborhood size, a time constant characterizes the neurons, a weight matrix, and input/output interconnection matrices. Some examples of the system behavior for different values of these parameters are presented to evaluate the suitability of the neural network prediction error model to fit a wide class of systems. The method is then applied to train the net to perform three classical problems: generation of Van der Pol oscillations, estimation of a sum of cosines in additive white Gaussian noise, and prediction of a chaotic time series generated by the logistic function.

AB - A class of neural-network based nonlinear models suitable for application in the prediction error systems identification method is presented. The parameters of this class of models include a local interconnection neighborhood size, a time constant characterizes the neurons, a weight matrix, and input/output interconnection matrices. Some examples of the system behavior for different values of these parameters are presented to evaluate the suitability of the neural network prediction error model to fit a wide class of systems. The method is then applied to train the net to perform three classical problems: generation of Van der Pol oscillations, estimation of a sum of cosines in additive white Gaussian noise, and prediction of a chaotic time series generated by the logistic function.

UR - http://www.scopus.com/inward/record.url?scp=0024807795&partnerID=8YFLogxK

U2 - 10.1109/acssc.1989.1200762

DO - 10.1109/acssc.1989.1200762

M3 - Conference contribution

AN - SCOPUS:0024807795

SN - 0929029301

T3 - Conference Record - Asilomar Conference on Circuits, Systems & Computers

SP - 119

EP - 122

BT - Conference Record - Asilomar Conference on Circuits, Systems & Computers

PB - Publ by Maple Press, Inc

T2 - Twenty-Third Annual Asilomar Conference on Signals, Systems &amp; Computers

Y2 - 30 October 1989 through 1 November 1989

ER -

A neural network model for prediction error identification

Abstract

Publication series

Conference

Access to Document

Fingerprint

Cite this