一种基于聚类思想的SVM多类分类方法

A Multi-Classification SVM Based on Clustering Idea

针对传统的基于决策树的支持向量机多类分类算法运算过程复杂、分类效率低的缺点,提出一种新的基于聚类思想的支持向量机分类方法.空间距离和聚类思想的引入,有效的提高了算法的分类效率.仿真试验表明,该方法在保持算法良好推广性的同时降低了算法的复杂度,从而提高了分类效率和分类速度.

Now that the original multi-classification SVM based on SVM-decision binary tree has low classification efficiency and its complicated operation,a multi-classification SVM based on clustering idea is presented in the paper.It improves the operation speed by employing the spatial distance and the clustering idea.The experimental results indicate that the new algorithm can not only keep the generalization capability but also improve the operation speed and classification efficiency.