Interactive Swarm Intelligence Algorithm Based on Master-Slave Gaussian Surrogate Model

An interactive swarm intelligence algorithm based on master-slave Gaussian surrogate model (ISIA-MSGSM) is proposed in this paper. In the algorithm, particle swarm optimization is used to act on the optimization search. During the search process, some data are sampled dynamically from the searching swarm to build the master and the slave Gaussian surrogate model, and all the particles will go through interactive evaluations based on the two kinds of surrogate models and the accurate model, which can reduce the computation cost of the objective function. At the same time, the surrogate models are managed dynamically guided by the accurate model to ensure the computational accuracy. Through the dynamical update to the master and slave model, the balance between the global exploration and the local exploitation is ensured which contributes to the efficiency of the algorithm. The experiment results on benchmark problems show this method not only can decrease the computation cost, but also has good robustness with a satisfied optimization performance.