Parallel Support Vector Machines

DSpace Repository

Show simple item record Brugger, Dominik de_DE 2007-03-08 de_DE 2014-03-18T10:16:32Z 2007-03-08 de_DE 2014-03-18T10:16:32Z 2006 de_DE
dc.identifier.other 286961369 de_DE
dc.identifier.uri de_DE
dc.description.abstract The Support Vector Machine (SVM) is a supervised algorithm for the solution of classification and regression problems. SVMs have gained widespread use in recent years because of successful applications like character recognition and the profound theoretical underpinnings concerning generalization performance. Yet, one of the remaining drawbacks of the SVM algorithm is its high computational demands during the training and testing phase. This article describes how to efficiently parallelize SVM training in order to cut down execution times. The parallelization technique employed is based on a decomposition approach, where the inner quadratic program (QP) is solved using Sequential Minimal Optimization (SMO). Thus all types of SVM formulations can be solved in parallel, including C-SVC and nu-SVC for classification as well as epsilon-SVR and nu-SVR for regression. Practical results show, that on most problems linear or even superlinear speedups can be attained. en
dc.language.iso en de_DE
dc.publisher Universität Tübingen de_DE
dc.rights ubt-podok de_DE
dc.rights.uri de_DE
dc.rights.uri en
dc.subject.classification Support-Vektor-Maschine , Parallelisierung , Maschinelles Lernen , Verteilte Programmierung , Quadratische Optimierung de_DE
dc.subject.ddc 620 de_DE
dc.subject.other Support Vector Machines , Machine Learning , Parallel Computing , Quadratic Optimization en
dc.title Parallel Support Vector Machines en
dc.type Report (Bericht) de_DE 2012-10-11 de_DE
utue.publikation.fachbereich Informatik de_DE
utue.publikation.fakultaet 7 Mathematisch-Naturwissenschaftliche Fakultät de_DE
dcterms.DCMIType Text de_DE
utue.publikation.typ report de_DE 2768 de_DE
utue.opus.portal wsi de_DE
utue.opus.portalzaehlung 2006.01000 de_DE
utue.publikation.source WSI ; 2006 ; 1 de_DE
utue.publikation.reihenname WSI-Reports - Schriftenreihe des Wilhelm-Schickard-Instituts für Informatik de_DE
utue.publikation.zsausgabe 2006, 1
utue.publikation.erstkatid 2919855-0


This item appears in the following Collection(s)

Show simple item record