Robust Graph-Based Multi-View Clustering

Graph-based multi-view clustering (G-MVC) constructs a graphical representation of each view and then fuses them to unified graph for clustering. Though demonstrating promising performance in various applications, we observe that their formulations are usually non-convex, leading local optimum. In this paper, propose novel MVC algorithm termed robust graph-based (RG-MVC) address issue. particular, define min-max formulation learning rewrite it as convex differentiable objective function whose convexity differentiability carefully proved. Thus, can efficiently solve the resultant problem using reduced gradient descent algorithm, corresponding solution is guaranteed be globally optimal. As consequence, although our free hyper-parameters, has shown good robustness against noisy views. Extensive experiments on benchmark datasets verify superiority proposed method compared state-of-the-art algorithms. Our codes appendix available at https://github.com/wx-liang/RG-MVC.