Monte Carlo Elites: Quality-Diversity Selection as a Multi-Armed Bandit Problem
This work addresses the problem of efficient quality-diversity search for evolutionary algorithms, offering an incremental improvement to MAP-Elites.
The paper tackled the challenge of balancing exploration and exploitation in evolutionary search by extending MAP-Elites to treat parent selection as a multi-armed bandit problem, using upper-confidence bounds to accelerate discovery and improve solution quality, resulting in more diverse and high-quality solutions across three testbeds.
A core challenge of evolutionary search is the need to balance between exploration of the search space and exploitation of highly fit regions. Quality-diversity search has explicitly walked this tightrope between a population's diversity and its quality. This paper extends a popular quality-diversity search algorithm, MAP-Elites, by treating the selection of parents as a multi-armed bandit problem. Using variations of the upper-confidence bound to select parents from under-explored but potentially rewarding areas of the search space can accelerate the discovery of new regions as well as improve its archive's total quality. The paper tests an indirect measure of quality for parent selection: the survival rate of a parent's offspring. Results show that maintaining a balance between exploration and exploitation leads to the most diverse and high-quality set of solutions in three different testbeds.