Improving RBF Kernel Function of Support Vector Machine using Particle Swarm Optimization

Support vector machine (SVM) has become an increasingly popular tool for machine learning tasks involving classification, regression or novelty detection. SVM is able to calculate the maximum margin (separating hyper-plane) between data with and without the outcome of interest if they are linearly separable. To improve the generalisation performance of SVM classifier optimization technique is used. Optimization refers to the selection of a best element from some set of available alternatives. Particle swarm optimization (PSO) is a population based stochastic optimization technique where the potential solutions, called particles, fly through the problem space by following the current optimum particles. In this paper, Principal Component Analysis (PCA) is used for reducing features of breast cancer, lung cancer and heart disease data sets and an empirical comparison of kernel selection using PSO for SVM is used to achieve better performance. This paper focused on SVM trained using linear, polynomial and radial basis function (RBF) kernels and applying PSO to each kernels for each data set to get better accuracy.