Community detection in complex networks using extended compact genetic algorithm

Complex networks are often studied as graphs, and detecting communities in a complex network can be modeled as a seriously nonlinear optimization problem. Soft computing techniques have shown promising results for solving this problem. Extended compact genetic algorithm (ECGA) use statistical learning mechanism to build a probability distribution model of all individuals in a population, and then create new population by sampling individuals according to their probability distribution instead of using traditional crossover and mutation operations. ECGA has distinct advantages in solving nonlinear and variable-coupled optimization problems. This paper attempts to apply ECGA to explore community structure in complex networks. Experimental results based on the GN benchmark networks, the LFR benchmark networks, and six real-world complex networks, show that ECGA is more effective than some other algorithms of community detection.