Incremental Multi-view Clustering

With the digital revolution, different issues arise, putting the community in real challenge front to the variety and the quantitative explosion of digital data. Thus, these data are increasingly collected from multiple sources or comprised of several views. The supplementary information describes distinct perspectives of data which can significantly improve the clustering performance and give more accurate results. In this work, we study the problem of multi-view clustering and we propose an incremental multi-view clustering method by considering all the views. The proposed method handles data with any number of views, taking into account the diversity of views. To get a robust result, we propose a fusion strategy of all views using L2-norm and unit variance to guarantee the consistence and the correctness between views. Then, we give an iterative solution allowing to divide the set of views into incremental subspaces. We compare our method with a number of methods proposed in the recent literature. The empirical evaluations on two real-world dataset demonstrate the effectiveness and the potential of our method.