Multiple kernel-based anchor graph coupled low-rank tensor learning for incomplete multi-view clustering

Abstract Incomplete Multi-View Clustering (IMVC) attempts to give an optimal clustering solution for incomplete multi-view data that suffer from missing instances in certain views. However, most existing IMVC methods still have various drawbacks practical applications, such as arbitrary scenarios cannot be handled; the computational cost is relatively high; valuable nonlinear relations among samples are often ignored; complementary information views not sufficiently exploited. To address above issues, this paper, we present a novel and flexible unified graph learning framework, called Multiple Kernel-based Anchor Graph coupled low-rank Tensor (MKAGT_IMVC), whose goal adaptively learn similarity matrix all Specifically, according characteristics of data, MKAGT_IMVC innovatively improves anchor selection strategy. Then, cross-view fusion mechanism introduced construct multiple fused complete graphs, which captures more intra-view inter-view relations. Moreover, model combining tensor constraint consensus developed, where graphs regarded prior knowledge initialize model. Extensive experiments conducted on eight datasets clearly show our method delivers superior performance relative some state-of-the-art terms ability time-consuming.