Abstract:

In order to construct an effective classifier for data streams with concept drifting,according to the theory that different data feature has different critical degree for classification,a method of Ensemble Classifier for Feature Drifting in data streams (ECFD) is proposed. Firstly,the definite of feature drifting and the relationship between feature drifting and concept drifting is given.Secondly,mutual information theory is used to propose an Unsupervised Feature Filter (UFF) technique,so that critical feature subsets are extracted to detect feature drifting.Finally, the basic classified algorithms with the capability of handling concept drifting is chosen to construct heterogeneous ensemble classifier on the basis of critical feature subsets. This method exhibits a new idea of way to highdimensional data streams with hidden concept drifting.Experimental results show that the method has strong appearance in accuracy, speed and scalability, especially for highdimensional data streams.
          

Key words: feature selection;feature drifting;concept drifting;data stream;mutual information;ensemble classifier