Forecasted Particle Swarm Optimization
Xingjuan Cai, Jianchao Zeng, Ying Tan · 2007
This paper introduces a novel fitness estimation strategy for particle swarm optimization (PSO) that does not evaluate all new positions, thus operating faster. A fitness and associated reliability value are assigned to each new individual that is only evaluated using the true fitness function if the reliability value is below some threshold. This variant of PSO designs a two-stage convex fitness estimation method. The first stage is used to estimate a visual position's fitness and reliability value, whereas in the second stage, the individual's fitness and reliability value are estimated with this visual position. Simulation results show the proposed algorithm is effective and efficient.