Correntropy based Low-rank Matrix Factorization with Multi-view Constraint Graph Learning for Multi-view Data Clustering

Abstract With the increase in data pattern diversity, multi-view clustering has become a hot topic for researchers. However, most of the current multi-view clustering methods focus on unsupervised learning scenarios, which cannot utilize the label information in the data. Furthermore, they could not handle the outliers, which may exist in the data. In order to address these issues, this paper proposes a Correntropy-based Multi-view Low-rank Matrix Factorization (CMLMF) method for multi-view data semi-supervised clustering. In order to sufficiently utilize the local structure information, a correntropy-based multi-view constraint graph learning framework is proposed to extract the local structure hidden in the multi-view data adaptively. Furthermore, a correntropy-based multi-view Low-rank Matrix Factorization (LMF) model is proposed, which can be incorporated with the adaptive graph learning framework to extract the global reconstruction information of data. The label information is introduced by a constraint matrix, which can be utilized for both global reconstruction and local structure information learning without adding any hyperparameter. Due to the maximum correntropy criterion, the proposed method is robust against the outliers in the data in each view and helps to extract a more current consensus representation. An efficient solving algorithm based on accelerated Block Coordinate Update (BCU) is designed to solve the model. Experiments are conducted on five real-world datasets, and the proposed method is compared with ten state-of-the-art methods to evaluate the effectiveness of the CMLMF method. The results show that the proposed method performs better clustering in most cases.