Selection of Gaussian kernel parameter for SVM based on convex estimation

Abstract

The selection of the kernel and the corresponding parameter is one of the key problems for support vector machine (SVM). This paper presents an approach to select the optimal Gaussian kernel parameter for support vector classification through the estimation of approximate convex hull of the sample set. The presented approach can overcome some disadvantages such as high computation cost existing in the traditional optimization-based methods, and it can be used no matter whether the dataset is dense or whether the distribution is uniform. The simulation experiments demonstrate the feasibility and the effectiveness of the presented method. © 2008 Springer-Verlag Berlin Heidelberg.