Summary
International Symposium on Nonlinear Theory and its Applications
2005
Session Number:1-2-3
Session:
Number:1-2-3-3
An Efficient Method for Searching Optimal Kernel Parameter of Support Vector Machines
Keisuke Arima, Norikazu Takahashi,
pp.238-241
Publication Date:2005/10/18
Online ISSN:2188-5079
Summary:
Generalization capability of support vector machines depends heavily on the kernel function and its parameters. In this paper, we focus our attention on the Gaussian kernel and propose an efficient method for finding the kernel parameter which minimizes the number of support vectors. Since the generalization error estimated by the leave-one-out procedure is upper-bounded by the ratio of the number of support vectors to the number of training samples, the proposed method will be useful for finding support vector machines with higher generalization capability.