Nonlinear System Identification With Composite Relevance Vector Machines

Author

Camps-Valls, Gustavo ; Martínez-Ramón, Manel ; Rojo-Álvarez, José Luis ; Muñoz-Marí, Jordi

Author_Institution

Dep. d´´Enginyeria Electronica, Valencia Univ.

Volume

14

Issue

4

fYear

2007

fDate

4/1/2007 12:00:00 AM

Firstpage

279

Lastpage

282

Abstract

Nonlinear system identification based on relevance vector machines (RVMs) has been traditionally addressed by stacking the input and/or output regressors and then performing standard RVM regression. This letter introduces a full family of composite kernels in order to integrate the input and output information in the mapping function efficiently and hence generalize the standard approach. An improved trade-off between accuracy and sparsity is obtained in several benchmark problems. Also, the RVM yields confidence intervals for the predictions, and it is less sensitive to free parameter selection

Keywords

nonlinear systems; regression analysis; support vector machines; composite kernels; composite relevance vector machine; mapping function; nonlinear system identification; standard RVM regression; Bayesian methods; Desktop publishing; Function approximation; Kernel; Nonlinear systems; Signal processing algorithms; Stacking; Support vector machine classification; Support vector machines; System identification; Composite kernels; nonlinear system identification; relevance vector machine (RVM);

fLanguage

English

Journal_Title

Signal Processing Letters, IEEE

Publisher

ieee

ISSN

1070-9908

Type

jour

DOI

10.1109/LSP.2006.885290

Filename

4130389