A bio-inspired genetic algorithm for community mining

Abstract

The community structure as a vital property for complex networks contributes a lot for understanding and detecting inherent functions of real networks. However, existing algorithms which are ranging from the optimization-based to model-based strategies still need to be strengthened further in terms of their robustness and accuracy. In this paper, a kind of multi-headed slime molds, Physarum, is used for optimizing genetic algorithm (GA), due to its intelligence of generating foraging networks based on bioresearches. Thus, a Physarum-based Network Model (PNM) is proposed based on the Physarum-based Model, which shows an ability of recognizing inter-community edges. Combining PNM with a genetic algorithm, a novel genetic algorithm, called PNGACD, is putting forward to enhance the GA's efficiency, in which a priori edge recognition of PNM is integrated into the phase of initialization. Moreover, experiments in six real-world networks are used to evaluate the efficiency of the proposed method. Results show that there is a remarkable improvement in term of the robustness and accuracy, which demonstrates that PNGACD has a better performance, compared with the existing algorithms.