A Multi-Agent Genetic Algorithm for Community Detection in Complex Networks

Zhangtao Li,Jing Liu

DOI: https://doi.org/10.1016/j.physa.2015.12.126

IF: 3.778

2016-01-01

Physica A Statistical Mechanics and its Applications

Abstract:Complex networks are popularly used to represent a lot of practical systems in the domains of biology and sociology, and the structure of community is one of the most important network attributes which has received an enormous amount of attention. Community detection is the process of discovering the community structure hidden in complex networks, and modularity Q is one of the best known quality functions measuring the quality of communities of networks. In this paper, a multi-agent genetic algorithm, named as MAGA-Net, is proposed to optimize modularity value for the community detection. An agent, coded by a division of a network, represents a candidate solution. All agents live in a lattice-like environment, with each agent fixed on a lattice point. A series of operators are designed, namely split and merging based neighborhood competition operator, hybrid neighborhood crossover, adaptive mutation and self-learning operator, to increase modularity value. In the experiments, the performance of MAGA-Net is validated on both well-known real-world benchmark networks and large-scale synthetic LFR networks with 5000 nodes. The systematic comparisons with GA-Net and Meme-Net show that MAGA-Net outperforms these two algorithms, and can detect communities with high speed, accuracy and stability.