Experimental Analysis of a Cooperative Coevolutionary Algorithm with Parameter Tuning for Multi-objective Problem Optimization with Uncertainty

Abstract

Currently, organizations face significant challenges demanding effective and efficient solutions. The problem optimization and decision-making coupled with Decision Maker Preferences (DMPs), are crucial for achieving success and maintaining a competitive edge. In many cases, business problems involve the need to optimize multiple conflicting objectives, and DMPs may not be entirely precise.Coevolutionary algorithms have become increasingly popular as effective tools for solving problems involving multiple objectives. These techniques enable the simultaneous evolution of multiple solutions through the interaction and joint improve of different populations. Coevolutionary algorithms promote cooperative solution improvement, fostering diversity and facilitating the discovery of optimal solutions to complex problems.Parameter tuning is critical in coevolutionary algorithms as it determines how potential solutions are explored and enhances their ability to avoid local optima, directing the search toward global solutions. In this article, an analysis is conducted to identify the most viable configurations using parameter tuning in a cooperative coevolutionary algorithm to solve multi-objective problems with uncertainty. Experimental results demonstrate that no configuration dominates by absolute distance, but options are identified that can generate high-quality solutions.