The effect of decoding fairness on particle swarm optimization for the p-median problem

Abstract

Solution encoding and decoding have a direct impact on meta-heuristic optimization methods. The mapping between search and solution spaces outlines the conditions for the metaheuristics and affects their ability to solve particular problems. The issue of encoding becomes especially pronounced when continuous metaheuristics are applied to discrete problems, in particular combinatorial problems with strict positional dependences in solution representations. This work takes a closer look at the decoding of combinations (fixed-length subsets) in continuous metaheuristics, demonstrates the inherent bias of simple combination decoding, and studies the effect of fair decoding in the context of the particle swarm optimization algorithm and the p-Median problem.