Indicator-Based Particle Swarm Optimization

Swarm intelligence algorithm is also one type of heuristic algortithms, just likegenetic algorithm. Early in1990s, rearchers had tried to develop stochastic algorithmsas simulations of simple social models from organisms. Researchers simulated socialbehavior of organisms through simplified assumption and developed two criticalalgorithms: ant colony algorithm(ACO) and particle swarm optimization (PSO).PSO is a stochastic optimization technique based on population and it stimulatesanimals’ social behavior. PSO has undergone many changes since its introduction in1995. As researchers have learned about the technique, they have derived newversions, developed new applications and published theorentical studies of the variousparameters and aspects of the algorithm.The development of indicator integrates user’s preference information intomultiobjective search. Based on indicators, a new PSO called Indicator-PSO isproposed in this paper. In Indicator-PSO, indicator is used for fitness assignment.Then we can find the best particle in the population and keep it in memory. We let itbe a parent and make it cross with other parents in the mating pool. So we takeadvantage of the best particle’s information and make the population move to a betterplace. The above measures make it easy to use PSO on multiobjective problems.Finally, the proposed Indicator-PSO algorithm is tested on five bi-objectivebenchmark test functions and experimental results demonstrate that Indicator-PSOoutperforms the two populer MOEAs NAGA-II and SPEA-2.