Adaptive Learning in Cartesian Product of Reproducing Kernel Hilbert Spaces

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

We propose a novel adaptive learning algorithm based on iterative orthogonal projections in the Cartesian product of multiple reproducing kernel Hilbert spaces (RKHSs). The objective is to estimate or track nonlinear functions that are supposed to contain multiple components such as i) linear and nonlinear components and ii) high-and low-frequency components. In this case, the use of multiple RKHSs permits a compact representation of multicomponent functions. The proposed algorithm is where two different methods of the author meet: multikernel adaptive filtering and the algorithm of hyperplane projection along affine subspace (HYPASS). In a particular case, the 'sum' space of the RKHSs is isomorphic, under a straightforward correspondence, to the product space, and hence the proposed algorithm can also be regarded as an iterative projection method in the sum space. The efficacy of the proposed algorithm is shown by numerical examples.

Original languageEnglish
Article number7174566
Pages (from-to)6037-6048
Number of pages12
JournalIEEE Transactions on Signal Processing
Volume63
Issue number22
DOIs
Publication statusPublished - 2015 Nov 15

Fingerprint

Hilbert spaces
Adaptive filtering
Adaptive algorithms
Learning algorithms

Keywords

  • Cartesian product
  • multikernel adaptive filtering
  • orthogonal projection
  • reproducing kernel Hilbert space

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this

Adaptive Learning in Cartesian Product of Reproducing Kernel Hilbert Spaces. / Yukawa, Masahiro.

In: IEEE Transactions on Signal Processing, Vol. 63, No. 22, 7174566, 15.11.2015, p. 6037-6048.

Research output: Contribution to journalArticle

@article{d592158b9de1463e87ee237f2c64d755,
title = "Adaptive Learning in Cartesian Product of Reproducing Kernel Hilbert Spaces",
abstract = "We propose a novel adaptive learning algorithm based on iterative orthogonal projections in the Cartesian product of multiple reproducing kernel Hilbert spaces (RKHSs). The objective is to estimate or track nonlinear functions that are supposed to contain multiple components such as i) linear and nonlinear components and ii) high-and low-frequency components. In this case, the use of multiple RKHSs permits a compact representation of multicomponent functions. The proposed algorithm is where two different methods of the author meet: multikernel adaptive filtering and the algorithm of hyperplane projection along affine subspace (HYPASS). In a particular case, the 'sum' space of the RKHSs is isomorphic, under a straightforward correspondence, to the product space, and hence the proposed algorithm can also be regarded as an iterative projection method in the sum space. The efficacy of the proposed algorithm is shown by numerical examples.",
keywords = "Cartesian product, multikernel adaptive filtering, orthogonal projection, reproducing kernel Hilbert space",
author = "Masahiro Yukawa",
year = "2015",
month = "11",
day = "15",
doi = "10.1109/TSP.2015.2463261",
language = "English",
volume = "63",
pages = "6037--6048",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "22",

}

TY - JOUR

T1 - Adaptive Learning in Cartesian Product of Reproducing Kernel Hilbert Spaces

AU - Yukawa, Masahiro

PY - 2015/11/15

Y1 - 2015/11/15

N2 - We propose a novel adaptive learning algorithm based on iterative orthogonal projections in the Cartesian product of multiple reproducing kernel Hilbert spaces (RKHSs). The objective is to estimate or track nonlinear functions that are supposed to contain multiple components such as i) linear and nonlinear components and ii) high-and low-frequency components. In this case, the use of multiple RKHSs permits a compact representation of multicomponent functions. The proposed algorithm is where two different methods of the author meet: multikernel adaptive filtering and the algorithm of hyperplane projection along affine subspace (HYPASS). In a particular case, the 'sum' space of the RKHSs is isomorphic, under a straightforward correspondence, to the product space, and hence the proposed algorithm can also be regarded as an iterative projection method in the sum space. The efficacy of the proposed algorithm is shown by numerical examples.

AB - We propose a novel adaptive learning algorithm based on iterative orthogonal projections in the Cartesian product of multiple reproducing kernel Hilbert spaces (RKHSs). The objective is to estimate or track nonlinear functions that are supposed to contain multiple components such as i) linear and nonlinear components and ii) high-and low-frequency components. In this case, the use of multiple RKHSs permits a compact representation of multicomponent functions. The proposed algorithm is where two different methods of the author meet: multikernel adaptive filtering and the algorithm of hyperplane projection along affine subspace (HYPASS). In a particular case, the 'sum' space of the RKHSs is isomorphic, under a straightforward correspondence, to the product space, and hence the proposed algorithm can also be regarded as an iterative projection method in the sum space. The efficacy of the proposed algorithm is shown by numerical examples.

KW - Cartesian product

KW - multikernel adaptive filtering

KW - orthogonal projection

KW - reproducing kernel Hilbert space

UR - http://www.scopus.com/inward/record.url?scp=84960158874&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84960158874&partnerID=8YFLogxK

U2 - 10.1109/TSP.2015.2463261

DO - 10.1109/TSP.2015.2463261

M3 - Article

AN - SCOPUS:84960158874

VL - 63

SP - 6037

EP - 6048

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 22

M1 - 7174566

ER -