Information entropy based interaction model and optimization method for swarm intelligence

Inspired by social insects, swarm intelligence has been hotly investigated in recent years as an innovative artificial intelligence technique for solving problems. In this paper, we mainly focus on the information interaction of individuals in swarm intelligence. By using information entropy H(X) and mutual information I(X;Y) of information theory to evaluate the information quality and interaction efficiency, respectively, the interaction model is proposed. Within this model, individuals' information is evaluated with uniform standards, so that more excellent individuals can be selected to influence other individuals by interaction. We validated this model with the route-exchange algorithm, which is proposed for combinatorial optimization. Seven benchmarks of the Traveling Salesman Problem are tested in the experiments. The results are compared with other heuristic algorithms.