Particle swarm optimization with average-fitness based selection
Stephen Y. Chen, Shanshan Lao, Irene Moser · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2022
The search trajectories of particles in Particle Swarm Optimization are influenced by attractions towards personal best positions. These personal best positions are meant to represent promising areas of the search space for further exploration and exploitation. However, the best position that has been visited by a particle may not be the best area for further exploration. Traditional fitness-based selection is suitable for identifying areas of the search space for further exploitation. Average-Fitness Based Selection is introduced as an alternative to identify the best areas for further exploration. Initial experiments demonstrate that this alternate form of selection can improve the performance of Particle Swarm Optimization in a multi-modal search space.