Are quality diversity algorithms better at generating stepping stones than objective-based search?

摘要: The route to the solution of complex design problems often lies through intermediate "stepping stones" which bear little resemblance final solution. By greedily following path greatest fitness improvement, objective-based search overlooks and discards stepping stones might be critical solving problem. Here, we hypothesize that Quality Diversity (QD) algorithms are a better way generate than search: by maintaining large set solutions high-quality, but phenotypically different, these collect promising while protecting them in their own "ecological niche". To demonstrate capabilities QD revisit challenge recreating images produced user-driven evolution, classic spurred work novelty illustrated limits search. We show far outperforms matching user-evolved images. Further, our results suggest some intriguing possibilities for leveraging diversity created QD.