Empirical study of the universum SVM learning for high-dimensional data

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

21 Scopus citations

Abstract

Many applications of machine learning involve sparse high-dimensional data, where the number of input features is (much) larger than the number of data samples, d≫n. Predictive modeling of such data is very ill-posed and prone to overfitting. Several recent studies for modeling high-dimensional data employ new learning methodology called Learning through Contradictions or Universum Learning due to Vapnik (1998,2006). This method incorporates a priori knowledge about application data, in the form of additional Universum samples, into the learning process. This paper investigates generalization properties of the Universum-SVM and how they are related to characteristics of the data. We describe practical conditions for evaluating the effectiveness of Random Averaging Universum.

Original languageEnglish (US)
Title of host publicationArtificial Neural Networks - ICANN 2009 - 19th International Conference, Proceedings
Pages932-941
Number of pages10
EditionPART 1
DOIs
StatePublished - 2009
Event19th International Conference on Artificial Neural Networks, ICANN 2009 - Limassol, Cyprus
Duration: Sep 14 2009Sep 17 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume5768 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other19th International Conference on Artificial Neural Networks, ICANN 2009
Country/TerritoryCyprus
CityLimassol
Period9/14/099/17/09

Fingerprint

Dive into the research topics of 'Empirical study of the universum SVM learning for high-dimensional data'. Together they form a unique fingerprint.

Cite this