UCL logo

UCL Discovery

UCL home » Library Services » Electronic resources » UCL Discovery

A kernel statistical test of independence

Gretton, A; Fukumizu, K; Teo, CH; Song, L; Schölkopf, B; Smola, AJ; (2009) A kernel statistical test of independence. In: UNSPECIFIED

Full text not available from this repository.

Abstract

Although kernel measures of independence have been widely applied in machine learning (notably in kernel ICA), there is as yet no method to determine whether they have detected statistically significant dependence. We provide a novel test of the independence hypothesis for one particular kernel independence measure, the Hilbert-Schmidt independence criterion (HSIC). The resulting test costs O(m 2 ), wherem is the sample size. We demonstrate that this test outperforms established contingency table and functional correlation-based tests, and that this advantage is greater for multivariate data. Finally, we show the HSIC test also applies to text (and to structured data more generally), for which no other independence test presently exists.

Type: Book chapter
Title: A kernel statistical test of independence
ISBN: 160560352X
URI: http://discovery.ucl.ac.uk/id/eprint/1334324
Downloads since deposit
0Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item