Weighted Least Squares Twin Large Margin Distribution Machine

Abstract

In order to improve training efficiency and generalization performance of twin support vector machine, a weighted least squares twin large margin distribution machine is proposed. In our approach, equality constraint technique is used to improve the training speed. The structural risk minimization principle is implemented by introducing a regularization term to improve classification accuracy. In addition, different weights are put on the error variables in order to eliminate the impact of noise data. The experimental results show that the proposed algorithm has better classification performance in testing accuracy and efficiency.