000146412 001__ 146412
000146412 005__ 20190316234731.0
000146412 037__ $$aREP_WORK
000146412 245__ $$aHierarchical Penalization
000146412 269__ $$a2007
000146412 260__ $$bIDIAP$$c2007
000146412 336__ $$aReports
000146412 500__ $$aTo appear in Advances in Neural Information Processing Systems 21 (NIPS 2007)
000146412 520__ $$aHierarchical penalization is a generic framework for incorporating prior information in the fitting of statistical models, when the explicative variables are organized in a hierarchical structure. The penalizer is a convex functional that performs soft selection at the group level, and shrinks variables within each group. This favors solutions with few leading terms in the final combination. The framework, originally derived for taking prior knowledge into account, is shown to be useful in linear regression, when several parameters are used to model the influence of one feature, or in kernel regression, for learning multiple kernels.
000146412 700__ $$aSzafranski, Marie
000146412 700__ $$aGrandvalet, Yves
000146412 700__ $$aMorizet-Mahoudeaux, Pierre
000146412 8564_ $$uhttp://publications.idiap.ch/downloads/reports/2007/grandvalet-idiap-rr-07-76.pdf$$zURL
000146412 8564_ $$s396278$$uhttps://infoscience.epfl.ch/record/146412/files/grandvalet-idiap-rr-07-76.pdf$$zn/a
000146412 909C0 $$0252189$$pLIDIAP$$xU10381
000146412 909CO $$ooai:infoscience.tind.io:146412$$pSTI$$preport$$qGLOBAL_SET
000146412 937__ $$aLIDIAP-REPORT-2007-049
000146412 970__ $$agrandvalet:rr07-76/LIDIAP
000146412 973__ $$aEPFL$$sPUBLISHED
000146412 980__ $$aREPORT