A twin projection support vector machine for data regression
详细信息 查看全文
- 作者：Xinjun Penga ; b ; xjpeng@shnu.edu.cn" class="auth_mailAuthor Vitae ; Dong XuaAuthor Vitae ; Jindong ShencAuthor Vitae
- 关键词：Support vector regression ; Projection axis ; Data structural information ; Empirical correlation coefficient ; Projection variance
- 刊名：Neurocomputing
- 出版年：22 August 2014
- 年：2014
- 卷：138
- 期：Complete
- 页码：131-141
- 全文大小：1061 K
文摘
- •
A twin projection support vector regression (TPSVR) is presented.
- •
TPSVR determines the regressor by two smaller sized SVM-type problems.
- •
TPSVR minimizes the variance of projected inputs.
- •
TPSVR maximizes the correlation coefficients between up- and down-bound targets and projected inputs.
- •
TPSVR obtains better generalization performance than other algorithms.