摘要
通过对统计学习理论中的支持向量回归问题,特别是结构风险问题和ε-不敏感函数的分析,得到了一种新的支持向量回归算法.新算法将传统的支持向量回归问题中的二次优化问题改进为线性规划问题,这一改进大大降低了求解的复杂度,其训练时间快了至少一个数量级以上.最后对人工和实际的样本进行了试验,结果说明了线性规划支持向量回归能较好地逼近被估计函数,且计算复杂度明显降低.
Based on analysis of the conclusions in the statistical learning theory, especially the structural risk minimization and the e-insensitive loss function, a novel linear programming support vector regression is proposed. The new algorithm reduces complexity and training time. Simulation results obtained using both artificial and real data show that the estimation performance of the proposed method is a good approximation of the unknown function and the computation complexity is significantly reduced,
出处
《上海大学学报(自然科学版)》
CAS
CSCD
北大核心
2006年第6期576-580,587,共6页
Journal of Shanghai University:Natural Science Edition
基金
上海市重点学科建设资助项目
国家高技术研究发展计划(863计划)资助项目(2002AA234021)
关键词
统计学习理论
结构风险
支持向量回归
线性规划
statistical learning theory
structural risk minimization
support vector regression
linear programming