Incomplete Multiview Clustering via Cross-View Relation Transfer

In this paper, we consider the problem of multi-view clustering on incomplete views. Compared with complete clustering, view-missing increases difficulty learning common representations from different To address challenge, propose a novel framework, which incorporates cross-view relation transfer and fusion learning. Specifically, based consistency existing in data, devise transfer-based completion module, transfers known similar inter-instance relationships to missing view infers data via graph networks transferred relationship graph. Then view-specific encoders are designed extract recovered an attention-based layer is introduced obtain representation. Moreover, reduce impact error caused by inconsistency between views better structure, joint optimize recovery simultaneously. Extensive experiments conducted several real datasets demonstrate effectiveness proposed method.