Sparse generalized kernel modeling for nonlinear systems

S. Chen; X. Hong; X. X. Wang; C. J. Harris

doi:10.1109/CDC.2005.1582550

Sparse generalized kernel modeling for nonlinear systems

S. Chen^*, X. Hong, X. X. Wang, C. J. Harris

^*Corresponding author for this work

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

A generalized kernel modeling approach is proposed for identification of discrete-time nonlinear systems. Each kernel regresser in the generalized kernel model has an individually fitted diagonal covariance matrix which is determined by maximizing the correlation between the regresser and training data. A state-of-the-art construction algorithm based on orthogonal least squares regression with leave-one-out test statistic and local regularization is applied to select a parsimonious generalized kernel model from the full regression matrix. The effectiveness of the proposed nonlinear modeling approach is demonstrated by the experimental results involving one simulated system and two real data sets.

Original language	English
Title of host publication	Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05
Pages	2574-2579
Number of pages	6
Volume	2005
DOIs	https://doi.org/10.1109/CDC.2005.1582550
Publication status	Published - 1 Dec 2005
Event	44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05 - Seville, United Kingdom Duration: 12 Dec 2005 → 15 Dec 2005

Conference

Conference	44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05
Country/Territory	United Kingdom
City	Seville
Period	12/12/05 → 15/12/05

Access to Document

10.1109/CDC.2005.1582550

Cite this

@inproceedings{8b174cdea83b4551b973fe7ac8efdf4e,

title = "Sparse generalized kernel modeling for nonlinear systems",

abstract = "A generalized kernel modeling approach is proposed for identification of discrete-time nonlinear systems. Each kernel regresser in the generalized kernel model has an individually fitted diagonal covariance matrix which is determined by maximizing the correlation between the regresser and training data. A state-of-the-art construction algorithm based on orthogonal least squares regression with leave-one-out test statistic and local regularization is applied to select a parsimonious generalized kernel model from the full regression matrix. The effectiveness of the proposed nonlinear modeling approach is demonstrated by the experimental results involving one simulated system and two real data sets.",

author = "S. Chen and X. Hong and Wang, {X. X.} and Harris, {C. J.}",

year = "2005",

month = dec,

day = "1",

doi = "10.1109/CDC.2005.1582550",

language = "English",

isbn = "0780395689",

volume = "2005",

pages = "2574--2579",

booktitle = "Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05",

note = "44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05 ; Conference date: 12-12-2005 Through 15-12-2005",

}

Chen, S, Hong, X, Wang, XX & Harris, CJ 2005, Sparse generalized kernel modeling for nonlinear systems. in Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05. vol. 2005, 1582550, pp. 2574-2579, 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05, Seville, United Kingdom, 12/12/05. https://doi.org/10.1109/CDC.2005.1582550

TY - GEN

T1 - Sparse generalized kernel modeling for nonlinear systems

AU - Chen, S.

AU - Hong, X.

AU - Wang, X. X.

AU - Harris, C. J.

PY - 2005/12/1

Y1 - 2005/12/1

N2 - A generalized kernel modeling approach is proposed for identification of discrete-time nonlinear systems. Each kernel regresser in the generalized kernel model has an individually fitted diagonal covariance matrix which is determined by maximizing the correlation between the regresser and training data. A state-of-the-art construction algorithm based on orthogonal least squares regression with leave-one-out test statistic and local regularization is applied to select a parsimonious generalized kernel model from the full regression matrix. The effectiveness of the proposed nonlinear modeling approach is demonstrated by the experimental results involving one simulated system and two real data sets.

AB - A generalized kernel modeling approach is proposed for identification of discrete-time nonlinear systems. Each kernel regresser in the generalized kernel model has an individually fitted diagonal covariance matrix which is determined by maximizing the correlation between the regresser and training data. A state-of-the-art construction algorithm based on orthogonal least squares regression with leave-one-out test statistic and local regularization is applied to select a parsimonious generalized kernel model from the full regression matrix. The effectiveness of the proposed nonlinear modeling approach is demonstrated by the experimental results involving one simulated system and two real data sets.

UR - http://www.scopus.com/inward/record.url?scp=33847185920&partnerID=8YFLogxK

U2 - 10.1109/CDC.2005.1582550

DO - 10.1109/CDC.2005.1582550

M3 - Conference publication

AN - SCOPUS:33847185920

SN - 0780395689

SN - 9780780395688

VL - 2005

SP - 2574

EP - 2579

BT - Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05

T2 - 44th IEEE Conference on Decision and Control, and the European Control Conference, CDC-ECC '05

Y2 - 12 December 2005 through 15 December 2005

ER -

Sparse generalized kernel modeling for nonlinear systems

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this