摘要
函数拟合通常是在有限的训练样本下对函数变量之间的关系做出预测,由于在实践中训练样本本身存在噪音和孤立点,用传统的方法进行函数拟合的效果不佳。考虑到不同特征对于回归问题相关程度的不同,研究了以灰色关联度作为权重的特征加权支持向量回归机算法,并推广运用于二维函数的回归拟合。仿真结果表明,灰色特征加权方法较传统支持向量回归机,具有更好的回归拟合能力。
Fitting function normally predict a relationship between variables by limited training samples of the trained function. In practice due to inherent noise and isolation of training samples, the results of fitting function often do not meet the require- ments by using traditional methods. Considering the difference of feature' s correlative degree to the regression problem becomes large, the support vector regression machine algorithm is researched, which uses the grey correlation grade as the feature weight. And extend in the application of the two-dimension functions fitting. The experiment resultal proves that this algorithm can get the better competence of regression fitting than the traditional support vector machine (SVR).
出处
《计算机工程与设计》
CSCD
北大核心
2012年第10期3975-3978,3983,共5页
Computer Engineering and Design
基金
重庆大学"211工程"三期创新人才培养计划建设基金项目(S-09110)
关键词
支持向量回归机
灰色关联度
特征
加权
拟合
support vector regression machine
grey correlation grade
feature
weighting
fitting