计算机应用 ›› 2011, Vol. 31 ›› Issue (02): 501-503.

• 人工智能 • 上一篇    下一篇

基于动量粒子群的混合核SVM参数优化方法

王佳1,徐蔚鸿2   

  1. 1. 长沙理工大学
    2. 长沙理工大学 计算机与通信工程学院
  • 收稿日期:2010-07-27 修回日期:2010-09-17 发布日期:2011-02-01 出版日期:2011-02-01
  • 通讯作者: 王佳
  • 基金资助:
    教育部重点科研基金项目;湖南省教育厅重点项目

Parameter optimization of mixed kernel SVM based on momentum particle swarm optimization

  • Received:2010-07-27 Revised:2010-09-17 Online:2011-02-01 Published:2011-02-01

摘要: 支持向量机(SVM)可以很好地用来解决分类问题,参数优化尤其重要。混合核函数的引入,使得SVM又多了一个可调参数。针对该参数用人工或经验的方法获取具有局限性,采用动量粒子群(MPSO)对SVM基本参数、混合可调核参数进行综合寻优,来寻找最佳参数组合。通过UCI数据仿真,对比结果表明:所提优化方法能够快速有效地提取最佳参数组合,所得SVM性能明显提高,分类效果更好。

关键词: 混合核, 动量粒子群优化, 参数优化, 分类

Abstract: Support Vector Machine (SVM) can be used to solve classification problems, and it is very important to optimize its parameters. With the introduction of mixed kernels, SVM has one more adjustable parameter. Because it is hard to obtain the parameter by manual or experience, Momentum Particle Swarm Optimization (MPSO) was used to find the best combination of the basic parameters and mixed adjustable nuclear parameter of SVM. Finally, the simulations of UCI data show that the proposed algorithm provides an effective way to search the best parameters combination, and makes SVM have higher performance and better classification accuracy.

Key words: mixed kernel, Momentum Particle Swarm Optimization (MPSO), parameter optimization, classification