关键词: 支持向量机, 混合核函数, 参数优化, 粒子群优化

Abstract: We proposed a hybrid kernel support vector machine (SVM) algorithm based on improved particle swarm  optimization (PSO) algorithm, which solved the problem that the general hybrid kernel SVM algorithm was difficult to evaluate the parameter selection. The algorithm improved the convergence property by limiting the velocity of the particle, the search space and the crossover operator to get the best combination of the parameters. Simulation experiments show that the algorithm can get the optimal value of parameters more quickly and effectively.

Key words: hybrid kernel function, particle swarm optimization （PSO）, parameter optimization, support vector machine (SVM)