Late Fusion Multi-view Clustering with Learned Consensus Similarity Matrix

Multiple kernel algorithms in a late fusion manner have been widely used because of its excellent performance and high efficiency in multi-view clustering (MVC). The existing MVC algorithms via late fusion obtain a consensus clustering indicator matrix through the linear combination of the base clustering indicator matrix. As a result, the optimal consensus indicator matrix’s searching space reduces, and the clustering effect is limited. To learn more information from the base clustering indicator matrices, we construct a consensus similarity matrix as the input of the spectral clustering algorithm. Furthermore, we design an effective iterative algorithm to solve the new resultant optimization problem. Extensive experiments on 11 multi-view benchmark datasets demonstrate the effectiveness and efficiency of the proposed algorithm.