Quality-diversity in dissimilarity spaces
arXiv:2211.12337 · doi:10.1145/3583131.3590409
Abstract
The theory of magnitude provides a mathematical framework for quantifying and maximizing diversity. We apply this framework to formulate quality-diversity algorithms in generic dissimilarity spaces. In particular, we instantiate and demonstrate a very general version of Go-Explore with promising performance.
Patched Section 7 (not in the GECCO 2023 version at DOI 10.1145/3583131.3590409) with inline forward reference to https://arxiv.org/html/2509.19565 and https://proceedings.mlr.press/v321/huntsman26a.html, which contain the correct algorithm and proofs. No experimental results are materially affected in either this version or the GECCO one
References in corpus (5)
- Graph Learning: A Survey
- Covariance Matrix Adaptation for the Rapid Illumination of Behavior Space
- Discovering Representations for Black-box Optimization
- Multi-Emitter MAP-Elites: Improving quality, diversity and convergence speed with heterogeneous sets of emitters
- Comparing reliability of grid-based Quality-Diversity algorithms using artificial landscapes