A Decomposition-Based Multiobjective Chemical Reaction Optimization Algorithm for Community Detection in Complex Networks

Abstract

Community detection structure is very important for understanding the organization of the complex networks. This problem is NP-hard, which is modeled as a seriously nonlinear optimization problem. Recently, different intelligence algorithm has shown promising results for this problem. The chemical reaction optimization (CRO) algorithm is a novel evolutionary algorithm which mimics the phenomenon of interactions among molecules in a container. The one characteristic of CRO is that the size of the population is changing. In this paper, we redefined the operator of CRO, and using the method of multiobjective decomposition decomposed the community detection problem into a scalar of sub-problems and using the proposed a discrete variant of CRO (MODCRO) to optimization. In the proposed method, neighbor-based turbulence of on-wall ineffective collision operator and decomposition operator are redefined which is responsible for searching local exploitation ability of algorithm, and the intermolecular ineffective collisions operator and synthesis operator is also redesigned which is responsible for searching global exploration ability of algorithm. Experimental results clearly demonstrate that the proposed algorithm outperforms a number of state-of-the-art multiobjective optimization evolutionary algorithms (MOEAs) on modularity.