Fast rates for support vector machines using Gaussian kernels

Simple item page
dc.creatorSteinwart, Ingo
dc.creatorScovel, Clint
dc.date2007-08-14
dc.date.accessioned2026-07-07T08:24:37Z
dc.date.available2026-07-07T08:24:37Z
dc.descriptionFor binary classification we establish learning rates up to the order of $n^{-1}$ for support vector machines (SVMs) with hinge loss and Gaussian RBF kernels. These rates are in terms of two assumptions on the considered distributions: Tsybakov's noise assumption to establish a small estimation error, and a new geometric noise condition which is used to bound the approximation error. Unlike previously proposed concepts for bounding the approximation error, the geometric noise assumption does not employ any smoothness assumption.
dc.descriptionPublished at http://dx.doi.org/10.1214/009053606000001226 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/0708.1838
dc.identifierhttp://arxiv.org/abs/0708.1838
dc.identifierAnnals of Statistics 2007, Vol. 35, No. 2, 575-607
dc.identifierdoi:10.1214/009053606000001226
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/136397
dc.subjectStatistics Theory
dc.subjectMachine Learning
dc.subject68Q32 (Primary); 62G20, 62G99, 68T05, 68T10, 41A46, 41A99 (Secondary)
dc.titleFast rates for support vector machines using Gaussian kernels
dc.typetext

Files

Collections

arXiv