Multi-objective Discrete Moth-Flame Optimization for Complex Network Clustering

Abstract

Complex network clustering has been extensively studied in recent years, mostly through optimization approaches. In such approaches, the multi-objective optimization methods have been shown to be capable of overcoming the limitations (e.g., instability) of the single-objective methods. Nevertheless, such methods suffer from the shortcoming of incapability of maintaining a good tradeoff between exploration and exploitation, that is, to find better solutions based on the good ones obtained so far. In this paper, we present a new nature-inspired heuristic optimization method, called multi-objective discrete moth-flame optimization (DMFO) method, which achieves such a tradeoff. We describe the detailed algorithm of DMFO that utilizes the Tchebycheff decomposition approach with an \(l_2\)-norm constraint on the direction vector (2-Tch). Furthermore, we show the experimental results on synthetic and several real-world networks that verify that the proposed DMFO and the algorithm are both effective and promising for tackling the task of complex network clustering.