Support Vector Machines with the Ramp Loss and the Hard Margin Loss

J. Paul Brooks ()
Additional contact information
J. Paul Brooks: Department of Statistical Sciences and Operations Research, Virginia Commonwealth University, Richmond, Virginia 23284

Operations Research, 2011, vol. 59, issue 2, 467-479

Abstract: In the interest of deriving classifiers that are robust to outlier observations, we present integer programming formulations of Vapnik's support vector machine (SVM) with the ramp loss and hard margin loss. The ramp loss allows a maximum error of 2 for each training observation, while the hard margin loss calculates error by counting the number of training observations that are in the margin or misclassified outside of the margin. SVM with these loss functions is shown to be a consistent estimator when used with certain kernel functions. In computational studies with simulated and real-world data, SVM with the robust loss functions ignores outlier observations effectively, providing an advantage over SVM with the traditional hinge loss when using the linear kernel. Despite the fact that training SVM with the robust loss functions requires the solution of a quadratic mixed-integer program (QMIP) and is NP-hard, while traditional SVM requires only the solution of a continuous quadratic program (QP), we are able to find good solutions and prove optimality for instances with up to 500 observations. Solution methods are presented for the new formulations that improve computational performance over industry-standard integer programming solvers alone.

Keywords: statistics; pattern analysis; support vector machines; programming; integer applications; quadratic integer programming (search for similar items in EconPapers)
Date: 2011
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (13)

Downloads: (external link)
http://dx.doi.org/10.1287/opre.1100.0854 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:inm:oropre:v:59:y:2011:i:2:p:467-479

Access Statistics for this article

More articles in Operations Research from INFORMS Contact information at EDIRC.
Bibliographic data for series maintained by Chris Asher ().