支持向量机的光滑与逼近关系研究

On Relationship between Smoothness and Approximation of Support Vector Machine

支持向量机是近年来数据挖掘领域发展起来的一个新方法.对现有的四个光滑支持向量机进行了分析,研究了支持向量机的光滑与逼近的关系.数值实验结果表明,支持向量机在具有二阶光滑的条件下,分类效果随逼近精度的提高而改善.

Support vector machine （SVM） is a new technique in data mining. This paper examines the four smooth SVMs and analyzes the relationship between smoothness and approximation. Numerical results show that with 2nd-order smoothness, SVM＇s classification effect increases with its approximation accuracy.