Auto-Weighted Multi-View Clustering for Large-Scale Data

Multi-view clustering has gained broad attention owing to its capacity exploit complementary information across multiple data views. Although existing methods demonstrate delightful performance, most of them are high time complexity and cannot handle large-scale data. Matrix factorization-based models a representative solving this problem. However, they assume that the views share dimension-fixed consensus coefficient matrix view-specific base matrices, limiting their representability. Moreover, series algorithms bear one or more hyperparameters impractical in real-world applications. To address two issues, we propose an auto-weighted multi-view (AWMVC) algorithm. Specifically, AWMVC first learns matrices from corresponding different dimensions, then fuses obtain optimal matrix. By mapping original features into distinctive low-dimensional spaces, can attain comprehensive knowledge, thus obtaining better results. design six-step alternative optimization algorithm proven be convergent theoretically. Also, shows excellent performance on various benchmark datasets compared with ones. The code is publicly available at https://github.com/wanxinhang/AAAI-2023-AWMVC.