Application of State Transition Simulated Annealing Algorithm in Community Detection

Community detection is one of the most important attributes to reveal the hidden structure of complex networks. The way of community detection based on the intelligent optimization algorithm has been widely used. Aiming at the initialization problem of the solution, the paper uses the density peaks clustering (DPC) algorithm to obtain a stable and high-quality solution. Aiming at the search strategy, state transition simulated annealing algorithm (STASA) is proposed. On the basis, three types of operators are designed: vertex replacement operator, community fusion operator and cross mutation are designed. In the initial optimization stage, the vertex substitution operator is used to generate diverse solution. The community fusion operator speeds up the optimization in a certain sense. The introduction of the cross mutation operator in the later optimization stage is only suitable for high-quality solution and enhances the local search ability of the algorithm. Finally, the experimental results on the GN benchmark and the real-world networks show that the algorithm is superior to other classic community detection algorithms in terms of stability and accuracy.