Let the Data Choose: Flexible and Diverse Anchor Graph Fusion for Scalable Multi-View Clustering

In the past few years, numerous multi-view graph clustering algorithms have been proposed to enhance performance by exploring information from multiple views. Despite superior performance, high time and space expenditures limit their scalability. Accordingly, anchor learning has introduced alleviate computational complexity. However, existing approaches can be further improved following considerations: (i) Existing anchor-based methods share same number of anchors across This strategy violates diversity flexibility data distribution. (ii) Searching for optimal within hyper-parameters takes much extra tuning time, which makes impractical. (iii) How flexibly fuse graphs diverse sizes not well explored in literature. To address above issues, we propose a novel method termed Flexible Diverse Anchor Graph Fusion Scalable Multi-view Clustering (FDAGF) this paper. Instead manually with massive hyper-parameters, optimize contribution weights group pre-defined numbers avoid expenditure among Most importantly, hybrid fusion multi-size theoretical proof, allows flexible fusion. Then, an efficient linear optimization algorithm is solve resultant problem. Comprehensive experimental results demonstrate effectiveness efficiency our framework. The source code available at https://github.com/Jeaninezpp/FDAGF.