A new support vector machine for the classification of positive and unlabeled examples
Junyan Tan, Ling Zhen, Nai-Yang Deng, Chunhua Zhang · 2013
In this paper, we propose a new version of support vector machine named biased p-norm support vector machine (BPSVM) involved in learning from positive and unlabeled examples. BPSVM treats the classification of positive and unlabeled examples as an imbalanced binary classification problem by giving different penalty parameters to positive and unlabeled examples. Compared with the previous works, BPSVM can not only improve the performance of classification but also select relevant features automatically. Furthermore, an effective algorithm for solving our new model is proposed. BPSVM can be used to solve large scale problem due to the effectiveness of the new algorithm. Numerical results show BPSVM is effective in both classification and features selection.