Evolutionary Parameter Estimation Algorithm for Combined Kernel Function in Support Vector Machine

Abstract

This paper proposes a new kernel function for support vector machine and its learning method with fast convergence and good classification performance. A set of kernel functions are combined to create a new kernel function, which is trained by a learning method based on evolution algorithm. The learning method results in the optimal decision model consisting of a set of features as well as a set of the parameters for combined kernel function. The combined kernel function and the learning method were applied to obtain the optimal decision model for the classification of clinical proteome patterns, and the combined kernel function showed faster convergence in learning phase and resulted in the optimal decision model with better classification performance than other kernel functions. Therefore, the combined kernel function has the greater flexibility in representing a problem space than single kernel functions.

This research was supported by IRC (Internet Information Retrieval Research Center) in Hankuk Aviation University. IRC is a Kyounggi-Province Regional Research Center designated by Korea Science and Engineering Foundation and Ministry of Science & Technology.