Multi-view Multi-label Learning with Incomplete Views and Labels

Data set with incomplete information, multi-granularity label correlation when label-specific features and complementarity information provided is ubiquitous in real-world applications. In this paper, we develop a new multi-view multi-label learning with incomplete views and labels (MVML-IVL) for solution and it is the first attempt to design a multi-view multi-label learning method with incomplete views and labels by the learning of label-specific features, label correlation matrix, low-rank assumption matrix, multi-granularity label correlation, and complementary information. Experimental results validate that (1) MVML-IVL achieves a better performance and it is superior to the classical multi-view (multi-label) learning methods in statistical; (2) the running time of MVML-IVL won’t add too much; (3) MVML-IVL has a good convergence and ability to process multi-view multi-label data sets; (4) multi-granularity label correlation plays an important role for the performance of MVML-IVL; (5) the influence of adjustable parameters is not too large.