Unified Embedding Alignment with Missing Views Inferring for Incomplete Multi-View Clustering

• DOI: 10.1609/aaai.v33i01.33015393
• Publication Date: 2019-08-30T07:33:20Z
• Authors: Jie Wen; Zheng Zhang; Yong Xu; Bob Zhang; Lunke Fei; Hong Liu
• Keywords: Regularization; Consensus clustering; Constrained clustering

Multi-view clustering aims to partition data collected from diverse sources based on the assumption that all views are complete. However, such prior is hardly satisfied in many real-world applications, resulting incomplete multi-view learning problem. The existing attempts this problem still have following limitations: 1) underlying semantic information of missing commonly ignored; 2) local structure not well explored; 3) importance different effectively evaluated. To address these issues, paper proposes a Unified Embedding Alignment Framework (UEAF) for robust clustering. In particular, locality-preserved reconstruction term introduced infer can be naturally aligned. A consensus graph adaptively learned and embedded via reverse regularization guarantee common multiple turn further align inferred views. Moreover, an adaptive weighting strategy designed capture Extensive experimental results show proposed method significantly improve performance comparison with some state-of-the-art methods.