Efficient Initialization of Particle Swarm Optimization Using Low Discrepancy Sequence

Particle Swarm Optimization is very popular metaheuristic used in optimization problems. The performance of PSO depends greatly on the way its parameter are set. The initialization strategy is one of such parameter. The performance of PSO can be fine-tuned by using an efficient initialization strategy. In this paper we use low discrepancy Sobol, and Halton sequences for initialization of PSO, and the results are compared with PSO initialized with uniform pseudorandom numbers. The experimental results show that the initialization with low discrepancy sequence can significantly improve the performance of PSO.