多元宇宙优化算法改进SVM参数

Based on multi-verse optimizer algorithm for SVM parameter optimization

针对支持向量机(SVM)参数难以选择和确定的问题,采用一种新式元启发式优化算法——多元宇宙优化算法(MVO).并在传统多元宇宙优化算法(MVO)的基础上针对TDR值下降速度慢而导致旅行距离增加的问题,提出改进多元宇宙优化算法(IMVO),将改进多元宇宙优化算法用于支持向量机的参数优化和选择问题上.使用UCI标准数据库中的数据进行数值仿真实验.研究结果表明:采用改进多元宇宙优化算法优化的支持向量机有较强的寻优性能,稳定性较好.

In order to solve the support vector machine （SVM） parameters is difficult to select and determine, this paper proposed a new meta-heuristie algorithm--Multi-verse optimizer algorithm （MVO）. For TDR reduced speed slow and cause travel distance increase, an improving multi-verse optimizer algorithm （IMVO） was proposed, which is used to optimize and choose the parameters of Support Vector Machine（SVM）, and the numerical simulation experiment was conducted with the datasets from University of Califomia Irvine（UCI）. The results show that the IMVO algorithm for SVM parameter optimization has stronger optimization performance and better stability.