A learning algorithm for one-class data stream classification based on ensemble classifier

Current research on data stream classification mainly focuses on supervised learning, in which a fully labeled data stream is needed for training. However, fully labeled data streams are expensive to obtain, which makes the supervised learning approach difficult to be applied to real-life applications. In this paper, we consider the problem of one-class classification on data stream with respect to concept drift where a large volume of data arrives at a high speed and with change of concept. In this case, only a small number of positively labeled examples is available for training. We propose our OcEC(One-class Ensemble Classifiers)algorithm and have it compared with WEC algorithm and SEA algorithm. Experimental study on both Moving Hyperplane dataset and SEA dataset shows that the OcEC algorithm has excellent classification performance and can quickly adapt to concept drift.