Local Rademacher complexities

Simple item page
dc.creatorBartlett, Peter L.
dc.creatorBousquet, Olivier
dc.creatorMendelson, Shahar
dc.date2005-08-16
dc.date.accessioned2026-07-07T08:07:08Z
dc.date.available2026-07-07T08:07:08Z
dc.descriptionWe propose new bounds on the error of learning algorithms in terms of a data-dependent notion of complexity. The estimates we establish give optimal rates and are based on a local and empirical version of Rademacher averages, in the sense that the Rademacher averages are computed from the data, on a subset of functions with small empirical error. We present some applications to classification and prediction with convex function classes, and with kernel classes in particular.
dc.descriptionPublished at http://dx.doi.org/10.1214/009053605000000282 in the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)
dc.identifierhttps://arxiv.org/abs/math/0508275
dc.identifierhttp://arxiv.org/abs/math/0508275
dc.identifierAnnals of Statistics 2005, Vol. 33, No. 4, 1497-1537
dc.identifierdoi:10.1214/009053605000000282
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/130840
dc.subjectStatistics Theory
dc.subject62G08, 68Q32 (Primary)
dc.titleLocal Rademacher complexities
dc.typetext

Files

Collections

arXiv