IJPAM: Volume 35, No. 2 (2007)

OPTIMAL CONDITIONS AND NUMERICAL
EXPERIMENTS IN PARAMETER SELECTION
PROBLEMS OF SVM

Yu-Lin Dong$^1$, Yun Wang$^2$, Zun-Quan Xia$^3$
$^{1,2,3}$Department of Applied Mathematics
Dalian University of Technology
Dalian, 116024, P.R. CHINA
$^{1}$e-mail: dyl@student.dlut.edu.cn
$^{2}$e-mail: wangyun_3412@163.com
$^{3}$e-mail: zqxiazhh@dlut.edu.cn


Abstract.Support vector machine (SVM) is a widely used tool for classification. In this paper, we present a new SVM model to calculate the optimal value of cost parameter $C$ for particular problems of linearity non-separability of data. The new SVM model is formulated in the form of one of MPEC problems with an integer objective function. A lower bound, positive number, $C_0$ is required to provide for avoiding choosing a candidate set of $C$. Numerical experiments show that this model for choice of $C$ is suitable for solving SVM problems.

Received: December 26, 2006

AMS Subject Classification: 90C90

Key Words and Phrases: data mining, support vector machine, cost parameter, nonlinear programming, MPEC problem

Source: International Journal of Pure and Applied Mathematics
ISSN: 1311-8080
Year: 2007
Volume: 35
Issue: 2