摘要
A new nonparallel twin support vector machine is presented in this article which combine l(2,1) -norm minimizing on both loss function and regularization which called (TWSVM_21). Our TWSVM_21 is formulated aiming to separate class more efficiently. Different from other nonparallel classifier, such as the presented TBSVM, the classifier we proposed in this paper joint l(2,1) -norm. Experiments on DDSM datasets show the feasibility and effectiveness of our TWSVM_21.
- 出版日期2016
- 单位武汉科技大学