Effective Incomplete Multi-View Clustering via Low-Rank Graph Tensor Completion

In the past decade, multi-view clustering has received a lot of attention due to popularity data. However, not all samples can be observed from every view some unavoidable factors, resulting in incomplete (IMC) problem. Up until now, most efforts for IMC problem have been made on learning consensus representations or graphs, while many missing views are ignored, making it impossible capture information hidden view. To overcome this drawback, we first analyzed low-rank relationship existing inside each graph and among then propose novel method via tensor completion. Specifically, stack similarity graphs into third-order exploit mode using matrix nuclear norm. way, connection between available instances recovered. The representation learned completed spectral clustering. obtain optimal result, recovery integrated joint framework optimization. Extensive experimental results several datasets demonstrate that proposed better performance comparison with state-of-the-art methods.