Manifold-regularized minimax probability machine

Kazuki Yoshiyama, Akito Sakurai

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

In this paper we propose Manifold-Regularized Minimax Probability Machine, called MRMPM. We show that Minimax Probability Machine can properly be extended to semi-supervised version in the manifold regularization framework and that its kernelized version is obtained for non-linear case. Our experiments show that the proposed methods achieve results competitive to existing learning methods, such as Laplacian Support Vector Machine and Laplacian Regularized Least Square for publicly available datasets from UCI machine learning repository.

Original languageEnglish
Title of host publicationPartially Supervised Learning - First IAPR TC3 Workshop, PSL 2011, Revised Selected Papers
Pages42-51
Number of pages10
DOIs
Publication statusPublished - 2012 Feb 28
Event1st IAPR-TC3 Workshop on Partially Supervised Learning, PSL 2011 - Ulm, Germany
Duration: 2011 Sep 152011 Sep 16

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7081 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st IAPR-TC3 Workshop on Partially Supervised Learning, PSL 2011
CountryGermany
CityUlm
Period11/9/1511/9/16

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Manifold-regularized minimax probability machine'. Together they form a unique fingerprint.

Cite this