关键词: 支持向量机, 参数选择, 遗传算法, 粒子群算法

Abstract:

Parameter selection is a very important factor to evaluating the performance of Support Vector Machine (SVM). SVM is helpful to solve the small sample, nonlinear problems, but is timeconsuming in solving large sample data sets and easy to fall into local optimal solution. Therefore, in order to reduce this shortage, this paper proposes to combine the genetic algorithm and the particle swarm optimizations to optimize parameter selection. Besides, we apply the model algorithm to the artificial experiment. The result shows that our proposal is a very efficient method, it can avoid falling into the partial solution and improve the convergence rate to shorten the optimization time.

Key words: support vector machine;parameter selection;genetic algorithm;particle swarm optimization