A Survey on Incomplete Multiview Clustering

Conventional multi-view clustering seeks to partition data into respective\ngroups based on the assumption that all views are fully observed. However, in\npractical applications, such as disease diagnosis, multimedia analysis, and\nrecommendation system, it is common to observe that not all views of samples\nare available in many cases, which leads to the failure of the conventional\nmulti-view clustering methods. Clustering on such incomplete multi-view data is\nreferred to as incomplete multi-view clustering. In view of the promising\napplication prospects, the research of incomplete multi-view clustering has\nnoticeable advances in recent years. However, there is no survey to summarize\nthe current progresses and point out the future research directions. To this\nend, we review the recent studies of incomplete multi-view clustering.\nImportantly, we provide some frameworks to unify the corresponding incomplete\nmulti-view clustering methods, and make an in-depth comparative analysis for\nsome representative methods from theoretical and experimental perspectives.\nFinally, some open problems in the incomplete multi-view clustering field are\noffered for researchers.\n