Particle swarm optimization with resets - Improving the balance between exploration and exploitation

Abstract

Exploration and exploitation are two important factors to consider in the design of optimization techniques. Two new techniques are introduced for particle swarm optimization: "resets" increase exploitation and "delayed updates" increase exploration. In general, the added exploitation with resets helps more with the lbest topology which is more explorative, and the added exploration with delayed updates helps more with the gbest topology which is more exploitive. © 2010 Springer-Verlag.