Parameter selection in SVM with RBF kernel function

Abstract

Kernel function parameter selection is one of the important parts of support vector machine (SVM) modeling. In this paper, we analyzed the features of double linear search method and the grid search method selection method features and the algorithm implementation steps, which consider the selection of RBF kernel function parameter as an example, based on the analysis it is also given the double linear grid search method, and we would get the selection of support vector machines (SVM) nuclear parameter of automatic transmission engineering vehicles by using this method. Experiments show, double linear grid search method sets the advantages which double linear search method of small amount of training and grid search method to learn high precision.