Twin support vector machines with privileged information

Abstract

In the field of machine learning, collected data always have additional features which are always referred as privileged information. Privileged information learning is mainly used to help train the classifier in the training process, and predict the unseen example by the learned classifier. In this paper, we propose a new method named twin support vector machines with privileged information (TWSVM-PI). In the proposed method, we first introduce the privileged information into twin SVMs so as to construct a model for prediction, and then utilize the Lagrangian multiplier method to optimize the proposed objective function. Thus, we obtain two nonparallel classification hyperplanes by solving two smaller sized quadratic programming problems (QPPs), which can shorter the computational time and improve the accuracy of the prediction. Finally, we conduct extensive experiments to evaluate the performance of the proposed TWSVM-PI method. The results have shown that our proposed method can obtain a better performance compared with state-of-the-art methods.