Twin Support Vector Machine With Universum

Universum is a new concept proposed recently, which is defined to be the sample that does not belong to any classes concerned. In this work, we propose a weighted Twin Support Vector Machine with Universum(called U-WTSVM) in this paper, where samples in the different positions are proposed to give different penalties. The final classifier can yield great generalization ability. Therefore U-WTSVM has better flexibility of the algorithm and can obtain more reasonable classifier in most case. All experiments demonstrate that U-WTSVM outperforms the traditional TSVM and U-TSVM.In this paper, we construct a new algorithm, called least squares projection twin support vector machine with Universum(U-LSPTSVM). Add Universum data is a very novel idea, Because it introduces sample has no relation with the sample of classification, And thereby introduces a priori domain information, so improve the classification performance of LSPTSVM algorithm. In addition, in order to promote its generalization ability, we also extend to recursive learning method used to further enhance the performance of U-LSPTSVM. Experiments show that, U-LSPTSVM can directly improve the training time of U-TSVM, and in most cases the experimental accuracy better than LSPTSVM.