Guiding Quality Diversity on Monotone Submodular Functions: Customising the Feature Space by Adding Boolean Conjunctions
Marcus Schmidbauer, Andre Opris, Jakob Bossek, Frank Neumann, Dirk Sudholt · Proceedings of the Genetic and Evolutionary Computation Conference · 2024
Quality Diversity (QD) aims to evolve a population of solutions that are both diverse and of high quality. The Map-Elites QD approach partitions the search space according to a feature space and stores the best solution for each feature. Bossek & Sudholt (GECCO 2023) showed that a simple QD algorithm on the feature space defined by the number of selected elements efficiently computes (1 - 1/e)-approximations for maximising monotone submodular functions.