Kernel weights for equalizing kernel-wise convergence rates of multikernel adaptive filtering

Kwangjin Jeong, Masahiro Yukawa

Research output: Contribution to journalArticlepeer-review

Abstract

Multikernel adaptive filtering is an attractive nonlinear approach to online estimation/tracking tasks. Despite its potential advantages over its single-kernel counterpart, a use of inappropriately weighted kernels may result in a negligible performance gain. In this paper, we propose an efficient recursive kernel weighting technique for multikernel adaptive filtering to activate all the kernels. The proposed weights equalize the convergence rates of all the corresponding partial coefficient errors. The proposed weights are implemented via a certain metric design based on the weighting matrix. Numerical examples show, for synthetic and multiple real datasets, that the proposed technique exhibits a better performance than the manually-tuned kernel weights, and that it significantly outperforms the online multiple kernel regression algorithm.

Original languageEnglish
Pages (from-to)927-939
Number of pages13
JournalIEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
Volume1
Issue number6
DOIs
Publication statusPublished - 2021 Jun 1

Keywords

  • Adaptive filtering
  • Kernel method
  • Online nonlinear estimation

ASJC Scopus subject areas

  • Signal Processing
  • Computer Graphics and Computer-Aided Design
  • Electrical and Electronic Engineering
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'Kernel weights for equalizing kernel-wise convergence rates of multikernel adaptive filtering'. Together they form a unique fingerprint.

Cite this