Active semi-supervised framework with data editing

Abstract

Self-labeled training data in semi-supervised learning may contain much noise due to the initial insufficient training data, which may hurt the generalization ability of the final hypothesis. In this paper, we propose an Active Semi-Supervised framework with Data Editing(ASSDE) to improve sparsely labeled text classification. A data editing technique is used to identify and remove noise introduced by semi-supervised labeling. We carry out the data editing technique by fully utilizing the advantage of active learning, which is novel according to our knowledge. The fusion of active learning with data editing makes ASSDE more robust to the sparseness and the distribution bias of the initial training data, and it further simplifies the design of semi-supervised learning which makes ASSDE more efficient. Extensive experimental study on several real-world text data sets shows the encouraging results of the proposed framework for sparsely labeled text classification, compared with several state-of-the-art methods.