自适应误差惩罚支撑向量回归机  

SVR with Adaptive Error Penalization

作　　者：陈晓峰[1] 王士同[1] 曹苏群[1] 

机构地区：[1]江南大学信息学院,无锡214122

出　　处：《电子与信息学报》2008年第2期367-370,共4页Journal of Electronics & Information Technology

基　　金：2004年教育部优秀人才支持计划(NCET-04-0496);模式识别国家重点实验室开放课题;南京大学软件新技术国家重点实验室开放课题;教育部重点科学研究项目(105087);国防应用基础研究基金项目(A1420061266)

摘　　要：该文提出一种支撑向量回归机AEPSVR。它首先用ε-SVR求得一个近似的支撑向量回归函数,在此基础上,引入一种新自适应误差惩罚函数,通过迭代,得到鲁棒的支撑向量回归机。该方法因以ε-SVR为基础,故可以应用各种求解SVR的优化算法。实验表明,该支撑向量回归机AEPSVR能显著地降低离群点的影响,具有良好的泛化性能。A novel support vector regression method AEPSVR is proposed in this paper. First, an approximate regression function is obtained using ε-SVR method, and then a new adaptive error penalization function is introduced to enhance the robust performance of SVR such that a robust support vector regression is derived. Because the proposed AEPSVR here is based on ε-SVR, so various optimization methods for SVR can be used. Experimental results show that the proposed AEPSVR can reduce the affect of outliers, and have the very good generalization capability.

关 键 词：支撑向量回归 离群点 自适应误差惩罚 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]