Multi-view Clustering via Co-regularized Nonnegative Matrix Factorization with Correlation Constraint

With the increasing availability of multi-view nonnegative data in practical applications, multi-view learning based on nonnegative matrix factorization (NMF) has attracted more and more attentions. However, previous works are either difficult to generate meaningful clustering results in terms of views with heterogeneous quality, or sensitive to noise. To address these problems, we propose a co-regularized nonnegative matrix factorization method with correlation constraint (CO-NMFCC) for multi-view clustering, which jointly exploits both consistent and complementary information across multiple views. Different from previous works, we aim at integrating information from multiple views efficiently and making it more robust to the presence of noisy views. More specifically, correlation constraint is imposed on the low-dimensional space to learn a common representation shared by multiple views. Meanwhile, we exploit the complementary information of multiple views through the coregularization to accommodate the imbalance of the quality of views. In addition, experiments on two real datasets demonstrate that CO-NMFCC is an effective and promising algorithm for practical applications.