Robust weighted co-clustering with global and local discrimination. (June 2023)

• Record Type: Journal Article
• Authors: Lu, Zhoumin; Wang, Shiping; Liu, Genggeng; Nie, Feiping
• Is Part Of: Pattern recognition. Volume 138(2023)
• Publication Date: 2023-06
• Subjects: Machine learning -- Co-clustering -- Nonnegative matrix factorization -- Global discrimination -- Local discrimination; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2023.109405
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Form a weighted 0-norm non-negative matrix tri-factorization with global and local discriminant regularizers to simultaneously considers feature weights, data noise, local manifolds, and global scatter in co-clustering. Optimize the formed objective function to solve the proposed co-clustering problem by an alternate update rule, whose convergence is proved in theory. Verify our algorithm's duality, robustness, effectiveness, and competitiveness through comprehensive experiments on synthetic, corrupted and real datasets. Abstract: In the past few decades, the clustering problem has made considerable progress, and co-clustering algorithms have attracted more attention. Compared with one-side clustering, co-clustering not only groups samples according to the distribution of features but also groups features according to the distribution of samples at the same time. This duality helps to explore the structural information of data, such as genes and texts. In this paper, a new co-clustering algorithm is proposed to simultaneously consider feature weights, data noise, local manifolds, and global scatter, named robust weighted co-clustering with global and local discrimination. Furthermore, an alternate update rule is put forward to optimize objective, theoretically proven to converge. Then, the algorithm's duality, robustness, and effectiveness have been verified on synthetic, corrupted, and real datasets, respectively. The runtime and parameter sensitivity of the algorithm are also analyzed. Finally, sufficient experiments clarify the competitiveness of our algorithm compared to other ones.