Cost sensitive active learning based on self-training

Abstract

In machine learning and data mining, in order to deal with the problem of combining labeled and unlabeled data to improve the classification performance, active learning has attracted much attention in recent years. However, most studies on active learning are cost-insensitive. Cost-sensitive learning is a type of learning that misclassification costs are taken into consideration in the learning algorithm. In this paper, we propose a cost-sensitive active learning algorithm based on self-training. In addition, labeling cost is also considered in this paper. The results of experiments show a better performance of our algorithm compared to the current methods.