A Novel Semi-Supervised SVM Based on Tri-Training

One of the main difficulties in machine learning is how to solve large-scale problems effectively, and the labeled data are limited and fairly expensive to obtain. In this paper a new semi-supervised SVM algorithm is proposed. It applies tri-training to improve SVM. The semi-supervised SVM makes use of the large number of unlabeled data to modify the classifiers iteratively. Although tri-training doesn't put any constraints on the classifier, the proposed method uses three different SVMs as the classification algorithm. Experiments on UCI datasets show that tri-training can improve the classification accuracy of SVM and can increase the difference of classifiers, the accuracy of final classifier will be higher. Theoretical analysis and experiments show that the proposed method has excellent accuracy and classification speed.