An adaptive differential evolution algorithm driven by multiple probabilistic mutation strategies for influence maximization in social networks

The influence maximization is regarded as one of the most pivotal concerns in social network analysis especially in the inevitable trend that more and more individuals are being involved into the global networked society. The purpose of the problem is to identify a set of influential nodes from the social network and activate them to maximize the expected number of influenced nodes at the end of the spreading process. Although some meta-heuristics based on swarm intelligence or biological evolution have been proposed to tackle this intractable problem, the exploration and exploitation operations need further investigated according to the iterative information of the evolutionary process. In this paper, an adaptive differential evolution algorithm driven by multiple probabilistic mutation strategies is proposed for the influence maximization problem. To provide more adequate seed set by utilizing the differential evolution, the mutation in the framework, consisted by three policies, namely particle swarm mutation strategy, differential mutation strategy, and perturbation strategy, is implied based on different probabilistic models. An adaptive local search strategy is presented to improve the local optimum based on a potential substitution pool consisting of structural hole nodes. Experimental results on six real-world social networks demonstrate that the proposed algorithm shows competitive performance in terms of both efficacy and efficiency when compared to the state-of-the-art algorithms.