支持向量回归机二阶充分条件理论研究

Theoretical study on second order sufficient conditions for support vector regression

支持向量回归机是数据挖掘的新方法。针对支持向量回归机所对应的回归问题给出了其解二阶充分条件成立的理论结果,结果表明支持向量回归机优化问题的解在一个很弱假设下即可满足二阶充分性;该理论为支持向量回归机优化问题解的灵敏度分析奠定了一定基础。

Support vector regression （SVR） is a new method for data mining. This paper presents one condition for second order sufficient property of support vector classification to hold. The condition is so weak that support vector classification meets it easily. The theory provides an important foundation for sensitivity analysis of SVR.