Value-at-risk support vector machine: Stability to outliers

Peter Tsyurmasto, Michael Zabarankin, Stan Uryasev

Research output: Contribution to journalArticlepeer-review

25 Scopus citations

Abstract

A support vector machine (SVM) stable to data outliers is proposed in three closely related formulations, and relationships between those formulations are established. The SVM is based on the value-at-risk (VaR) measure, which discards a specified percentage of data viewed as outliers (extreme samples), and is referred to as VaR -SVM. Computational experiments show that compared to the ν -SVM, the VaR-SVM has a superior out-of-sample performance on datasets with outliers.

Original languageEnglish
Pages (from-to)218-232
Number of pages15
JournalJournal of Combinatorial Optimization
Volume28
Issue number1
DOIs
StatePublished - Jul 2014

Keywords

  • Classification
  • Conditional value-at-risk
  • Optimization
  • Risk management
  • Support vector machine
  • Value-at-risk

Fingerprint

Dive into the research topics of 'Value-at-risk support vector machine: Stability to outliers'. Together they form a unique fingerprint.

Cite this