Between-class discriminant twin support vector machine for imbalanced data classification

Most techniques for dealing with imbalanced data classification in SVM-related methods are sampling, weighting and ensemble strategies, which aim at balancing the importance of different classes when computing the decision boundary. But how to choose and design an appropriate balance strategy is still a difficult problem. In this paper, we propose a between-class discriminant twin support vector machine (BDTSVM) for imbalanced data classification. Between-class scatter matrix is insensitive to the imbalanced distributions of classes, so we introduce it as a regularization term into twin support vector machine to enhance its discriminant ability. In this way, BDTSVM wouldn't ignore the minority class. Moreover, the proposed BDTSVM can be solved by efficient eigenvalue decomposition, which avoids computing inversion matrix and quadratic programming problem as in TWSVM. Experimental results show, our BDTSVM achieves higher accuracy and G-mean on artificial and benchmark UCI datasets, compared with traditional strategies based SVM and TWSVM.