一种基于类中心最大间隔的支持向量机

A Support Vector Machine Based on Maximal Class-Center Margin

传统的支持向量机分类超平面对噪声和野值非常敏感.使用传统的支持向量机对含有噪声的数据分类时,所得到的超平面往往不是最优超平面.为了解决这个问题,本文以两个类中心距离最大为准则建立分类超平面,构造一个新的支持向量机,称作类中心最大间隔支持向量机.理论分析和仿真实验结果证明了该方法的正确性和有效性.

The separating hyperplane of traditional support vector machines is sensitive to noises and outliers. When traditional support vector machines separate data containing noises, the obtained hyperplane is not an optimal one. For this problem, a separating hyperplane is designed with the principle of maximizing the distance between two class centers, and a novel support vector machine, called maximal class-center margin support vector machine （ MC- CM-SVM） is designed. Theoretical analysis and experimental results show that the presented method is correct and effective.