Presentation 2011-11-09
Approximate Reduction from AUC Maximization to 1-norm Soft Margin Optimization
Daiki SUEHIRO, Kohei HATANO, Eiji TAKIMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Finding linear classifiers that maximize AUC scores is important in ranking research. This is naturally formulated as a 1-norm hard/soft margin optimization problem over pn pairs of p positive and n negative instances. However, directly solving the optimization problems is impractical since the problem size (pn) is quadratically larger than the given sample size (p+n). In this paper, we give (approximate) reductions from the problems to hard/soft margin optimization problems of linear size. First, for the hard margin case, we show that the problem is reduced to a hard margin optimization problem over p+n instances in which the bias constant term is to be optimized. Then, for the soft margin case, we show that the problem is approximately reduced to a soft margin optimization problem over p+n instances for which the resulting linear classifier is guaranteed to have a certain margin over pairs.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) ranking / AUC / soft margin optimization / reduction
Paper # IBISML2011-48
Date of Issue

Conference Information
Committee IBISML
Conference Date 2011/11/2(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Information-Based Induction Sciences and Machine Learning (IBISML)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Approximate Reduction from AUC Maximization to 1-norm Soft Margin Optimization
Sub Title (in English)
Keyword(1) ranking
Keyword(2) AUC
Keyword(3) soft margin optimization
Keyword(4) reduction
1st Author's Name Daiki SUEHIRO
1st Author's Affiliation Department of Informatics, Kyushu University()
2nd Author's Name Kohei HATANO
2nd Author's Affiliation Department of Informatics, Kyushu University
3rd Author's Name Eiji TAKIMOTO
3rd Author's Affiliation Department of Informatics, Kyushu University
Date 2011-11-09
Paper # IBISML2011-48
Volume (vol) vol.111
Number (no) 275
Page pp.pp.-
#Pages 8
Date of Issue