Domain Adaptation with Twin Support Vector Machines

摘要

Supervised learning models assume that the training and test data are drawn from the same underlying distribution. These algorithms become ineffective when such assumptions are violated. Therefore, domain adaptation methods were proposed to handle data from different, yet correlated distributions. They have been applied in many fields such as computer vision, sentiment analysis, natural language processing and so on. Often in these cases, source domains provide large numbers of labeled data for learning, while target domain may have few labeled data available. In this paper, we explore a new supervised domain adaption method with twin support vector machines called aggregate domain adaptation twin support vector machines, which obtain the classifiers trained on the source and target training data. Then we evaluate the performance of our proposed methods on brain-computer interfaces and sentiment analysis datasets. The experimental results show the effectiveness and the reliability of our proposed methods.