Kernel-based methods for hypothesis testing: A unified view

Zaid Harchaoui; Francis Bach; Olivier Cappe; Eric Moulines

doi:10.1109/MSP.2013.2253631

Kernel-based methods for hypothesis testing: A unified view

Zaid Harchaoui
, Francis Bach
, Olivier Cappe
, Eric Moulines

INRIA Institut National de Recherche en Informatique et en Automatique
PSL research University & IPSL
Telecom Paris
CNRS LTCI

Research output: Contribution to journal › Article › peer-review

Abstract

Kernel-based methods provide a rich and elegant framework for developing nonparametric detection procedures for signal processing. Several recently proposed procedures can be simply described using basic concepts of reproducing kernel Hilbert space (RKHS) embeddings of probability distributions, mainly mean elements and covariance operators. We propose a unified view of these tools and draw relationships with information divergences between distributions.

Original language	English
Article number	6530767
Pages (from-to)	87-97
Number of pages	11
Journal	IEEE Signal Processing Magazine
Volume	30
Issue number	4
DOIs	https://doi.org/10.1109/MSP.2013.2253631
Publication status	Published - 1 Jan 2013
Externally published	Yes

Access to Document

10.1109/MSP.2013.2253631

Cite this

@article{bda818468a3e4ae8b3a42279d46cbaf0,

title = "Kernel-based methods for hypothesis testing: A unified view",

abstract = "Kernel-based methods provide a rich and elegant framework for developing nonparametric detection procedures for signal processing. Several recently proposed procedures can be simply described using basic concepts of reproducing kernel Hilbert space (RKHS) embeddings of probability distributions, mainly mean elements and covariance operators. We propose a unified view of these tools and draw relationships with information divergences between distributions.",

author = "Zaid Harchaoui and Francis Bach and Olivier Cappe and Eric Moulines",

year = "2013",

month = jan,

day = "1",

doi = "10.1109/MSP.2013.2253631",

language = "English",

volume = "30",

pages = "87--97",

journal = "IEEE Signal Processing Magazine",

issn = "1053-5888",

number = "4",

}

TY - JOUR

T1 - Kernel-based methods for hypothesis testing

T2 - A unified view

AU - Harchaoui, Zaid

AU - Bach, Francis

AU - Cappe, Olivier

AU - Moulines, Eric

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Kernel-based methods provide a rich and elegant framework for developing nonparametric detection procedures for signal processing. Several recently proposed procedures can be simply described using basic concepts of reproducing kernel Hilbert space (RKHS) embeddings of probability distributions, mainly mean elements and covariance operators. We propose a unified view of these tools and draw relationships with information divergences between distributions.

AB - Kernel-based methods provide a rich and elegant framework for developing nonparametric detection procedures for signal processing. Several recently proposed procedures can be simply described using basic concepts of reproducing kernel Hilbert space (RKHS) embeddings of probability distributions, mainly mean elements and covariance operators. We propose a unified view of these tools and draw relationships with information divergences between distributions.

U2 - 10.1109/MSP.2013.2253631

DO - 10.1109/MSP.2013.2253631

M3 - Article

AN - SCOPUS:85032752162

SN - 1053-5888

VL - 30

SP - 87

EP - 97

JO - IEEE Signal Processing Magazine

JF - IEEE Signal Processing Magazine

IS - 4

M1 - 6530767

ER -

Kernel-based methods for hypothesis testing: A unified view

Abstract

Access to Document

Fingerprint

Cite this