{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T22:24:27Z","timestamp":1792794267266,"version":"4.5.0"},"reference-count":50,"publisher":"Oxford University Press (OUP)","issue":"1","license":[{"start":{"date-parts":[[2024,9,28]],"date-time":"2024-09-28T00:00:00Z","timestamp":1727481600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/pages\/standard-publication-reuse-rights"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,1,16]]},"abstract":"<jats:title>Summary<\/jats:title>\n               <jats:p>The kernel two-sample test based on the maximum mean discrepancy is one of the most popular methods for detecting differences between two distributions over general metric spaces. In this paper we propose a method to boost the power of the kernel test by combining maximum mean discrepancy estimates over multiple kernels using their Mahalanobis distance. We derive the asymptotic null distribution of the proposed test statistic and use a multiplier bootstrap approach to efficiently compute the rejection region. The resulting test is universally consistent and, since it is obtained by aggregating over a collection of kernels\/bandwidths, is more powerful in detecting a wide range of alternatives in finite samples. We also derive the distribution of the test statistic for both fixed and local contiguous alternatives. The latter, in particular, implies that the proposed test is statistically efficient, that is, it has nontrivial asymptotic (Pitman) efficiency. The consistency properties of the Mahalanobis and other natural aggregation methods are also explored when the number of kernels is allowed to grow with the sample size. Extensive numerical experiments are performed on both synthetic and real-world datasets to illustrate the efficacy of the proposed method over single-kernel tests. The computational complexity of the proposed method is also studied, both theoretically and in simulations. Our asymptotic results rely on deriving the joint distribution of the maximum mean discrepancy estimates using the framework of multiple stochastic integrals, which is more broadly useful, specifically, in understanding the efficiency properties of recently proposed adaptive maximum mean discrepancy tests based on kernel aggregation and also in developing more computationally efficient, linear-time tests that combine multiple kernels. We conclude with an application of the Mahalanobis aggregation method for kernels with diverging scaling parameters.<\/jats:p>","DOI":"10.1093\/biomet\/asae048","type":"journal-article","created":{"date-parts":[[2024,9,28]],"date-time":"2024-09-28T17:35:06Z","timestamp":1727544906000},"source":"Crossref","is-referenced-by-count":3,"title":["Boosting the power of kernel two-sample tests"],"prefix":"10.1093","volume":"112","author":[{"given":"A","family":"Chatterjee","sequence":"first","affiliation":[{"name":"Department of Statistics and Data Science, The Wharton School, University of Pennsylvania , 265 37th Street , Philadelphia, Pennsylvania 19104,","place":["USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"B B","family":"Bhattacharya","sequence":"additional","affiliation":[{"name":"Department of Statistics and Data Science, The Wharton School, University of Pennsylvania , 265 37th Street , Philadelphia, Pennsylvania 19104,","place":["USA"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2024,9,28]]},"reference":[{"key":"2025041509254542000_asae048-B1","doi-asserted-by":"crossref","first-page":"1148","DOI":"10.1214\/aoms\/1177704477","article-title":"On the distribution of the two-sample Cramer-von Mises criterion","volume":"33","author":"Anderson","year":"1962","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B2","doi-asserted-by":"crossref","first-page":"337","DOI":"10.1090\/S0002-9947-1950-0051437-7","article-title":"Theory of reproducing kernels","volume":"68","author":"Aronszajn","year":"1950","journal-title":"Trans. Am. Math. Soc"},{"key":"2025041509254542000_asae048-B3","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1080\/00949650410001661440","article-title":"New test for the multivariate two-sample problem based on the concept of minimum energy","volume":"75","author":"Aslan","year":"2005","journal-title":"J. Statist. Comp. Simul"},{"key":"2025041509254542000_asae048-B4","author":"Banerjee","year":"2022"},{"key":"2025041509254542000_asae048-B5","doi-asserted-by":"crossref","first-page":"190","DOI":"10.1016\/S0047-259X(03)00079-4","article-title":"On a new multivariate two-sample test","volume":"88","author":"Baringhaus","year":"2004","journal-title":"J. Mult. Anal"},{"key":"2025041509254542000_asae048-B6","doi-asserted-by":"crossref","first-page":"855","DOI":"10.1007\/s11063-016-9556-4","article-title":"Learning sparse feature representations using probabilistic quadtrees and deep belief nets","volume":"45","author":"Basu","year":"2017","journal-title":"Neural Proces. Lett"},{"key":"2025041509254542000_asae048-B7","doi-asserted-by":"crossref","first-page":"575","DOI":"10.1111\/rssb.12319","article-title":"A general asymptotic framework for distribution-free graph-based two-sample tests","volume":"81","author":"Bhattacharya","year":"2019","journal-title":"J. R. Statist. Soc. B"},{"key":"2025041509254542000_asae048-B8","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1214\/aoms\/1177697800","article-title":"A distribution free version of the Smirnov two sample test in the p-variate case","volume":"40","author":"Bickel","year":"1969","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B9","doi-asserted-by":"crossref","first-page":"913","DOI":"10.1093\/biomet\/asu045","article-title":"A distribution-free two-sample run test applicable to high-dimensional data","volume":"101","author":"Biswas","year":"2014","journal-title":"Biometrika"},{"key":"2025041509254542000_asae048-B10","doi-asserted-by":"crossref","first-page":"397","DOI":"10.1080\/01621459.2016.1147356","article-title":"A new graph-based two-sample test for multivariate and object data","volume":"112","author":"Chen","year":"2017","journal-title":"J. Am. Statist. Assoc"},{"key":"2025041509254542000_asae048-B11","first-page":"1981","author":"Chwialkowski","year":"2015"},{"key":"2025041509254542000_asae048-B12","doi-asserted-by":"crossref","first-page":"192","DOI":"10.1080\/01621459.2021.1923508","article-title":"Multivariate rank-based distribution-free nonparametric testing using measure transportation","volume":"118","author":"Deb","year":"2021","journal-title":"J. Am. Statist. Assoc"},{"key":"2025041509254542000_asae048-B13","doi-asserted-by":"crossref","first-page":"697","DOI":"10.1214\/aos\/1176344722","article-title":"Multivariate generalizations of the Wald-Wolfowitz and Smirnov two-sample tests","volume":"7","author":"Friedman","year":"1979","journal-title":"Ann. Statist."},{"key":"2025041509254542000_asae048-B14","doi-asserted-by":"crossref","first-page":"1431","DOI":"10.1214\/13-AOS1114","article-title":"The two-sample problem for poisson processes: adaptive tests with a nonasymptotic wild bootstrap approach","volume":"41","author":"Fromont","year":"2013","journal-title":"Ann. Statist."},{"key":"2025041509254542000_asae048-B15","first-page":"1","author":"Fromont","year":"2012"},{"key":"2025041509254542000_asae048-B16","first-page":"723","article-title":"A kernel two-sample test","volume":"13","author":"Gretton","year":"2012","journal-title":"J. Mach. Learn. Res"},{"key":"2025041509254542000_asae048-B17","first-page":"673","author":"Gretton","year":"2009"},{"key":"2025041509254542000_asae048-B18","first-page":"1205","author":"Gretton","year":"2012"},{"key":"2025041509254542000_asae048-B19","doi-asserted-by":"crossref","first-page":"359","DOI":"10.1093\/biomet\/89.2.359","article-title":"Permutation tests for equality of distributions in high-dimensional settings","volume":"89","author":"Hall","year":"2002","journal-title":"Biometrika"},{"key":"2025041509254542000_asae048-B20","doi-asserted-by":"crossref","first-page":"259","DOI":"10.1007\/BF01915210","article-title":"On the number of random points with nearest neighbour of the same type and a multivariate two-sample test","volume":"31","author":"Henze","year":"1984","journal-title":"Metrika"},{"key":"2025041509254542000_asae048-B21","doi-asserted-by":"crossref","first-page":"3417","DOI":"10.1214\/19-AOS1936","article-title":"Robust multivariate nonparametric tests via projection averaging","volume":"48","author":"Kim","year":"2020","journal-title":"Ann. Statist."},{"key":"2025041509254542000_asae048-B22","doi-asserted-by":"crossref","first-page":"411","DOI":"10.1214\/20-AOS1962","article-title":"Classification accuracy as a proxy for two-sample testing","volume":"49","author":"Kim","year":"2021","journal-title":"Ann. Statist."},{"key":"2025041509254542000_asae048-B23","first-page":"6245","author":"K\u00fcbler","year":"2020"},{"key":"2025041509254542000_asae048-B24","volume-title":"Testing Statistical Hypotheses","author":"Lehmann","year":"2005","edition":"3rd ed"},{"key":"2025041509254542000_asae048-B25","author":"Li","year":"2019"},{"key":"2025041509254542000_asae048-B26","first-page":"6316","author":"Liu","year":"2020"},{"key":"2025041509254542000_asae048-B27","doi-asserted-by":"crossref","first-page":"252","DOI":"10.1080\/01621459.1993.10594317","article-title":"A quality index based on data depth and multivariate rank tests","volume":"88","author":"Liu","year":"1993","journal-title":"J. Am. Statist. Assoc"},{"key":"2025041509254542000_asae048-B28","first-page":"1895","author":"Lopez-Paz","year":"2017"},{"key":"2025041509254542000_asae048-B29","doi-asserted-by":"crossref","first-page":"50","DOI":"10.1214\/aoms\/1177730491","article-title":"On a test of whether one of two random variables is stochastically larger than the other","volume":"18","author":"Mann","year":"1947","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B30","doi-asserted-by":"crossref","first-page":"1109","DOI":"10.1214\/17-AOS1579","article-title":"Ball divergence: nonparametric two sample test","volume":"46","author":"Pan","year":"2018","journal-title":"Ann. Statist"},{"key":"2025041509254542000_asae048-B31","first-page":"21247","author":"Park","year":"2020"},{"key":"2025041509254542000_asae048-B32","doi-asserted-by":"crossref","first-page":"47","DOI":"10.3390\/e19020047","article-title":"On Wasserstein two-sample testing and related families of nonparametric tests","volume":"19","author":"Ramdas","year":"2017","journal-title":"Entropy"},{"key":"2025041509254542000_asae048-B33","author":"Ramdas","year":"2015"},{"key":"2025041509254542000_asae048-B34","author":"R Development Core Team","year":"2024"},{"key":"2025041509254542000_asae048-B35","volume-title":"Methods of Modern Mathematical Physics. I: Functional Analysis","author":"Reed","year":"1980"},{"key":"2025041509254542000_asae048-B36","doi-asserted-by":"crossref","first-page":"515","DOI":"10.1111\/j.1467-9868.2005.00513.x","article-title":"An exact distribution-free test comparing two multivariate distributions based on adjacency","volume":"67","author":"Rosenbaum","year":"2005","journal-title":"J. R. Statist. Soc. B"},{"key":"2025041509254542000_asae048-B37","doi-asserted-by":"crossref","first-page":"799","DOI":"10.1080\/01621459.1986.10478337","article-title":"Multivariate two-sample tests based on nearest neighbors","volume":"81","author":"Schilling","year":"1986","journal-title":"J. Am. Statist. Assoc"},{"key":"2025041509254542000_asae048-B38","first-page":"1","article-title":"MMD aggregated two-sample test","volume":"24","author":"Schrab","year":"2023","journal-title":"J. Mach. Learn. Res"},{"key":"2025041509254542000_asae048-B39","doi-asserted-by":"crossref","first-page":"2263","DOI":"10.1214\/13-AOS1140","article-title":"Equivalence of distance-based and RKHS-based statistics in hypothesis testing","volume":"41","author":"Sejdinovic","year":"2013","journal-title":"Ann. Statist."},{"key":"2025041509254542000_asae048-B40","author":"Shekhar","year":"2022"},{"key":"2025041509254542000_asae048-B41","doi-asserted-by":"crossref","first-page":"279","DOI":"10.1214\/aoms\/1177730256","article-title":"Table for estimating the goodness of fit of empirical distributions","volume":"19","author":"Smirnov","year":"1948","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B42","author":"Song","year":"2023"},{"key":"2025041509254542000_asae048-B43","author":"Sutherland","year":"2021"},{"key":"2025041509254542000_asae048-B44","first-page":"1249","article-title":"Testing for equal distributions in high dimension","volume":"5","author":"Sz\u00e9kely","year":"2004","journal-title":"InterStat"},{"key":"2025041509254542000_asae048-B45","doi-asserted-by":"crossref","first-page":"1249","DOI":"10.1016\/j.jspi.2013.03.018","article-title":"Energy statistics: a class of statistics based on distances","volume":"143","author":"Sz\u00e9kely","year":"2013","journal-title":"J. Statist. Plan. Infer"},{"key":"2025041509254542000_asae048-B46","doi-asserted-by":"crossref","first-page":"791","DOI":"10.1093\/biomet\/asaa027","article-title":"Combining p-values via averaging","volume":"107","author":"Vovk","year":"2020","journal-title":"Biometrika"},{"key":"2025041509254542000_asae048-B47","doi-asserted-by":"crossref","first-page":"147","DOI":"10.1214\/aoms\/1177731909","article-title":"On a test whether two samples are from the same population","volume":"11","author":"Wald","year":"1940","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B48","doi-asserted-by":"crossref","first-page":"159","DOI":"10.1214\/aoms\/1177705995","article-title":"Two-sample tests for multivariate distributions","volume":"31","author":"Weiss","year":"1960","journal-title":"Ann. Math. Statist."},{"key":"2025041509254542000_asae048-B49","doi-asserted-by":"crossref","first-page":"119","DOI":"10.2307\/3001946","article-title":"Probability tables for individual comparisons by ranking methods","volume":"3","author":"Wilcoxon","year":"1947","journal-title":"Biometrics"},{"key":"2025041509254542000_asae048-B50","doi-asserted-by":"crossref","first-page":"105286","DOI":"10.1016\/j.jeconom.2022.03.007","article-title":"Testing equality of several distributions in separable metric spaces: a maximum mean discrepancy based approach","volume":"239","author":"Zhang","year":"2024","journal-title":"J. Economet"}],"container-title":["Biometrika"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/biomet\/advance-article-pdf\/doi\/10.1093\/biomet\/asae048\/60926609\/asae048.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/112\/1\/asae048\/60926609\/asae048.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/biomet\/article-pdf\/112\/1\/asae048\/60926609\/asae048.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,4,15]],"date-time":"2025-04-15T09:25:56Z","timestamp":1744709156000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/biomet\/article\/doi\/10.1093\/biomet\/asae048\/7789470"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,9,28]]},"references-count":50,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2025,1,16]]}},"URL":"https:\/\/doi.org\/10.1093\/biomet\/asae048","relation":{},"ISSN":["1464-3510"],"issn-type":[{"value":"1464-3510","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2025]]},"published":{"date-parts":[[2024,9,28]]},"article-number":"asae048"}}