One-step Incomplete Multi-view Clustering based on Bipartite Graph Learning

Although previous graph-based multi-view clustering algorithms have made remarkable progress, most of them still face the following two limitations: 1. Many existing methods rely on k-means for the discretization of spectral embeddings, which cannot directly learn graphs with discrete cluster structures and require two steps for clustering results. 2. Practical applications may contain some missing instances, which require Incomplete Multi-View Clustering (IMVC) methods to hold them. In this paper, we propose a novel method named One-step Incomplete Multi-View Clustering based on Bipartite Graph Learning (OIMVC-BGL) which aims to solve the above problems. OIMVC-BGL first constructs bipartite graphs from all views with an anchor-based subspace learning method. Then, OIMVC-BGL fuses these graphs to obtain a consensus bipartite graph with an adaptive weight manner. Finally, OIMVC-BGL imposes a Laplacian rank constraint on the consensus bipartite graph to obtain the results directly. Experiments conducted on benchmark datasets verify the effectiveness of OIMVC-BGL.