Localized Sparse Incomplete Multi-View Clustering

Incomplete multi-view clustering, which aims to solve the clustering problem\non the incomplete multi-view data with partial view missing, has received more\nand more attention in recent years. Although numerous methods have been\ndeveloped, most of the methods either cannot flexibly handle the incomplete\nmulti-view data with arbitrary missing views or do not consider the negative\nfactor of information imbalance among views. Moreover, some methods do not\nfully explore the local structure of all incomplete views. To tackle these\nproblems, this paper proposes a simple but effective method, named localized\nsparse incomplete multi-view clustering (LSIMVC). Different from the existing\nmethods, LSIMVC intends to learn a sparse and structured consensus latent\nrepresentation from the incomplete multi-view data by optimizing a sparse\nregularized and novel graph embedded multi-view matrix factorization model.\nSpecifically, in such a novel model based on the matrix factorization, a l1\nnorm based sparse constraint is introduced to obtain the sparse low-dimensional\nindividual representations and the sparse consensus representation. Moreover, a\nnovel local graph embedding term is introduced to learn the structured\nconsensus representation. Different from the existing works, our local graph\nembedding term aggregates the graph embedding task and consensus representation\nlearning task into a concise term. Furthermore, to reduce the imbalance factor\nof incomplete multi-view learning, an adaptive weighted learning scheme is\nintroduced to LSIMVC. Finally, an efficient optimization strategy is given to\nsolve the optimization problem of our proposed model. Comprehensive\nexperimental results performed on six incomplete multi-view databases verify\nthat the performance of our LSIMVC is superior to the state-of-the-art IMC\napproaches. The code is available in https://github.com/justsmart/LSIMVC.\n