用混沌理论优化SVM核参数的研究

混沌优化算法是一种有效的全局优化算法,其计算复杂度较低,搜索速度快。支持向量机是近年来新兴的模式识别方法,在解决小样本、非线性及高维模式识别问题中表现出了突出的优点。但支持向量机的识别性能对于参数的选择是敏感的,提出用混沌优化算法来优化支持向量机的参数,不仅提高了支持向量机的性能,而且解决了传统的选取参数方法计算量大、参数多时难以奏效的问题。仿真结果表明性能较好、计算量较少。

Optimizing the Parameters of SVM with Chaotic Optimization Algorithm

The chaos optimization algorithm can get global solution with low computational load. Support Vector Machine, a novel method d the pattern recognition, presents excellent pedormance in solving the problems with small sample, nonlinear and local mivima in recent years. However, SVM is sensitive to the parameters and traditional methods are not effective, especially, more parameters. In this paper, a chaotic optimization algorithm is presented for Support Vector Machine with Gaussian kernel so that the problems can be solved. The simulation results show that this algorithm is feasible and effective.