Comparing Surrogate-Assisted Evolutionary Algorithms on Optimization of a Simulation Model for Resource Planning Task for Hospitals

Surrogate-assisted evolutionary algorithms (SAEAs) are currently among the most widely researched techniques for their capability to solve expensive real-world optimization problems. The development of these techniques and their bench-marking with other methods still relies almost exclusively on artificially created problems. In this paper, we use a real-world problem of optimizing the parameters of a hospital resource planning tool to compare the performance of nine state-of-the-art single-objective SAEAs. We find that there are significant differences between the performance of the compared methods on the selected instances, making the problems suitable for benchmarking SAEAs.