Svm Performance Optimization Using PSO for Breast Cancer Classification

Abstract

Breast cancer is a very serious disease and requires the sufferer to undergo an intensive examination. This study proposes an optimization method for SVM parameters using the particle swarm optimization algorithm to classify breast cancer. The test results show that the classification using SVM with optimization using PSO is able to improve accuracy better than the classification using SVM without optimization, namely the determination of the parameters randomly. Breast cancer data classification accuracy increased to 78.91%. The best parameter values for c, γ, r, and d are 0.252101, 0.053248, 1 and 5, respectively. In this study, the polynomial and RBF kernels in general were able to produce higher accuracy than linear and sigmoid kernels.