Supervised feature selection via dependence estimation

Le Song; Alex Smola; Arthur Gretton; Karsten M. Borgwardt; Justin Bedo

doi:10.1145/1273496.1273600

Supervised feature selection via dependence estimation

Le Song^*, Alex Smola, Arthur Gretton, Karsten M. Borgwardt, Justin Bedo

^*Corresponding author for this work

Research output: Contribution to conference › Paper › peer-review

291 Citations (Scopus)

Abstract

We introduce a framework for filtering features that employs the Hilbert-Schmidt Independence Criterion (HSIC) as a measure of dependence between the features and the labels. The key idea is that good features should maximise such dependence. Feature selection for various supervised learning problems (including classification and regression) is unified under this framework, and the solutions can be approximated using a backward-elimination algorithm. We demonstrate the usefulness of our method on both artificial and real world datasets.

Original language	English
Pages	823-830
Number of pages	8
DOIs	https://doi.org/10.1145/1273496.1273600
Publication status	Published - 2007
Externally published	Yes
Event	24th International Conference on Machine Learning, ICML 2007 - Corvalis, OR, United States Duration: 20 Jun 2007 → 24 Jun 2007

Conference

Conference	24th International Conference on Machine Learning, ICML 2007
Country/Territory	United States
City	Corvalis, OR
Period	20/06/07 → 24/06/07

Access to Document

10.1145/1273496.1273600

Cite this

@conference{5db636149702487f828d2eba6bb2666c,

title = "Supervised feature selection via dependence estimation",

abstract = "We introduce a framework for filtering features that employs the Hilbert-Schmidt Independence Criterion (HSIC) as a measure of dependence between the features and the labels. The key idea is that good features should maximise such dependence. Feature selection for various supervised learning problems (including classification and regression) is unified under this framework, and the solutions can be approximated using a backward-elimination algorithm. We demonstrate the usefulness of our method on both artificial and real world datasets.",

author = "Le Song and Alex Smola and Arthur Gretton and Borgwardt, {Karsten M.} and Justin Bedo",

year = "2007",

doi = "10.1145/1273496.1273600",

language = "English",

pages = "823--830",

note = "24th International Conference on Machine Learning, ICML 2007 ; Conference date: 20-06-2007 Through 24-06-2007",

}

TY - CONF

T1 - Supervised feature selection via dependence estimation

AU - Song, Le

AU - Smola, Alex

AU - Gretton, Arthur

AU - Borgwardt, Karsten M.

AU - Bedo, Justin

PY - 2007

Y1 - 2007

N2 - We introduce a framework for filtering features that employs the Hilbert-Schmidt Independence Criterion (HSIC) as a measure of dependence between the features and the labels. The key idea is that good features should maximise such dependence. Feature selection for various supervised learning problems (including classification and regression) is unified under this framework, and the solutions can be approximated using a backward-elimination algorithm. We demonstrate the usefulness of our method on both artificial and real world datasets.

AB - We introduce a framework for filtering features that employs the Hilbert-Schmidt Independence Criterion (HSIC) as a measure of dependence between the features and the labels. The key idea is that good features should maximise such dependence. Feature selection for various supervised learning problems (including classification and regression) is unified under this framework, and the solutions can be approximated using a backward-elimination algorithm. We demonstrate the usefulness of our method on both artificial and real world datasets.

UR - http://www.scopus.com/inward/record.url?scp=34547964410&partnerID=8YFLogxK

U2 - 10.1145/1273496.1273600

DO - 10.1145/1273496.1273600

M3 - Paper

AN - SCOPUS:34547964410

SP - 823

EP - 830

T2 - 24th International Conference on Machine Learning, ICML 2007

Y2 - 20 June 2007 through 24 June 2007

ER -

Supervised feature selection via dependence estimation

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this