Staying together maybe better for particles

In nature, staying together is often of great selective advantage for social animals. Social animals frequently make consensus decisions, not least about group movements, in order to maintain group cohesion. Inspired by this social behavior, this paper proposes a new Particle Swarm Optimizer Based on Group Decision-Making (PSOGDM). Unlike the existing variants of PSO, historical information, such as gbest and pbest, are abandoned in PSOGDM. Instead, a consensus is decided by some elitists in the group using their current position information to lead the group members. All group members search towards the same consensus, as well as this memoryless consensus also encourages the swarm to jump out the local optima. The algorithm is experimentally validated on 20 benchmark functions. Experimental results show that the new algorithm performs much better than three popular PSO variants. Furthermore, compared with three well-know evolutionary algorithms, the results empirically demonstrate that the proposed algorithm also yields promising search performance.