Multi-View Clustering on Topological Manifold

Multi-view clustering has received a lot of attentions in data mining recently. Though plenty works have been investigated on this topic, it is still severe challenge due to the complex nature multiple heterogeneous features. Particularly, existing multi-view algorithms fail consider topological structure data, which essential for manifold. In paper, we propose exploit implied manifold by learning relationship between points. Our method coalesces view-wise graphs with relevance considered, and learns weights as well consensus graph interactively unified framework. Furthermore, manipulate connectivity constraint such that points from same cluster are precisely connected into component. Substantial experiments both toy real datasets conducted validate effectiveness proposed method, compared state-of-the-art over performance.