Artificial Intelligence for Metaheuristic Parameter Setting

The scientific content of this paper focuses on service system optimization. The primary objective of the research being presented consists in extending current portfolio of solving approaches for such discrete location problems, where two competing goals need to be optimized. Due to different optimization criteria, only a tiny portion of feasible solutions respecting a specific non-dominance property must be looked for. Acquiring the entire Pareto set is a time-consuming challenge. This occurrence has prompted the creation of various metaheuristic approaches, which are able to bring an approximation of the original Pareto frontier. Mentioned approximate methods may be affected by different parameters. That is why we pay attention to artificial intelligence for metaheuristic parameter settings. Suggested algorithm was examined on middle-sized real world benchmarks. The obtained Pareto front approximations are compared to the complete sets of non-dominated solutions, which are available.