WCSE

Link Selection with Collaborative Low-rank and Sparse Factorization for Community Detection in Multiplex Networks

Abstract— With the advent of multiple types of proximities between nodes, multiplex networks have emerged widely in the real world and been attracting increasing attention recently. The existing researches on community detection in multiplex networks usually assume that all layers come from a same latent consistent topology structure, and learn compatible and complementary information from different layers together, so as to dig out a shared community structure. However, this assumption is not satisfied in many real-world networks due to the existence of noisy/irrelevant links. To address this problem, in this paper we propose a multiplex network structure optimization algorithm based on collaborative low-rank and sparse factorization, which promotes the collaboration of different layers and lets them decompose as the robust consistent representations. In addition, an effective iterative algorithm is designed to optimize and solve the model. The experimental results in multiple ground-truth datasets show that this method can significantly improve the community detection performance of multiplex complex networks

Index Terms— Multiplex networks, Low-rank sparse representation, Link Selection, Community detection

Dengdi Sun
School of Computer Science and Technology, Anhui University, CHINA
Xinxin Wang
School of Computer Science and Technology, Anhui University, CHINA
Zhuanlian Ding
School of Internet, Anhui University, CHINA
Bin Luo
School of Computer Science and Technology, Anhui University, CHINA

[Download]

Cite: Dengdi Sun, Xinxin Wang, Zhuanlian Ding, Bin Luo , "Link Selection with Collaborative Low-rank and Sparse Factorization for Community Detection in Multiplex Networks ," 2021 The 11th International Workshop on Computer Science and Engineering (WCSE 2021), pp. 284-291, Shanghai, China, June 19-21, 2021.