Advanced Search
DSpace@MIT

Bagging Regularizes

Research and Teaching Output of the MIT Community

Show simple item record

dc.contributor.author Poggio, Tomaso en_US
dc.contributor.author Rifkin, Ryan en_US
dc.contributor.author Mukherjee, Sayan en_US
dc.contributor.author Rakhlin, Alex en_US
dc.date.accessioned 2004-10-20T21:04:57Z
dc.date.available 2004-10-20T21:04:57Z
dc.date.issued 2002-03-01 en_US
dc.identifier.other AIM-2002-003 en_US
dc.identifier.other CBCL-214 en_US
dc.identifier.uri http://hdl.handle.net/1721.1/7268
dc.description.abstract Intuitively, we expect that averaging --- or bagging --- different regressors with low correlation should smooth their behavior and be somewhat similar to regularization. In this note we make this intuition precise. Using an almost classical definition of stability, we prove that a certain form of averaging provides generalization bounds with a rate of convergence of the same order as Tikhonov regularization --- similar to fashionable RKHS-based learning algorithms. en_US
dc.format.extent 7 p. en_US
dc.format.extent 906324 bytes
dc.format.extent 285651 bytes
dc.format.mimetype application/postscript
dc.format.mimetype application/pdf
dc.language.iso en_US
dc.relation.ispartofseries AIM-2002-003 en_US
dc.relation.ispartofseries CBCL-214 en_US
dc.subject AI en_US
dc.subject Bagging en_US
dc.subject stability en_US
dc.subject regularization en_US
dc.title Bagging Regularizes en_US


Files in this item

Name Size Format Description
AIM-2002-003.ps 885.0Kb Postscript
AIM-2002-003.pdf 278.9Kb PDF

This item appears in the following Collection(s)

Show simple item record

MIT-Mirage