Output details
11 - Computer Science and Informatics
University College London
Return to search
Output 0 of 0 in the submission
Article title
A Kernel Two-Sample Test
Type
D - Journal article
DOI
-
Title of journal
Journal of Machine Learning Research
Article number
-
Volume number
13
Issue number
-
First page of article
723
ISSN of journal
1532-4435
Year of publication
2012
URL
-
Number of additional authors
4
Additional information
<13>Two-sample testing for high dimensional variables is difficult, and (prior to the conference version from 2007, cited 250 times) no two-sample tests existed for strings (eg. documents) and graphs (eg. proteins). This work provides the first two sample test for graphs, strings, and groups, as well as a multivariate test. The journal publication contributes a theoretical foundation for kernel tests: the first proofs of the asymptotic distribution, and of consistency against local departures from the null. A topic of numerous invited talks in the last two years, including an ICML2012 workshop and APRM2012 conference, Google, Yahoo, Microsoft, and Winton Capital.
Interdisciplinary
-
Cross-referral requested
-
Research group
None
Citation count
13
Proposed double-weighted
No
Double-weighted statement
-
Reserve for a double-weighted output
No
Non-English
No
English abstract
-
