On the use of feature-maps for improved quality-diversity meta-evolution

Bossens, David and Tarapore, Danesh (2021) On the use of feature-maps for improved quality-diversity meta-evolution. In GECCO '21: Proceedings of the Genetic and Evolutionary Computation Conference Companion. pp. 83-84 . (doi:10.1145/3449726.3459442).

Record type: Conference or Workshop Item (Paper)

Abstract

Quality-Diversity (QD) algorithms evolve a behaviourally diverse archive of high-performing solutions. In QD meta-evolution, one evolves a population of QD algorithms by modifying algorithmic components (e.g., the behaviour space) to optimise an archive-level objective, the meta-fitness. This paper investigates which feature-map is best for defining the behaviour space for an 8-joint robot arm. Meta-evolution with non-linear feature-maps yields a 15-fold meta-fitness improvement over linear feature-maps. On a damage recovery test, archives evolved with non-linear feature-maps outperform traditional MAP-Elites variants.

This record has no associated files available for download.

More information

Published date: 7 July 2021

Additional Information: Funding Information: This work was supported by EPSRC under the New Investigator Award grant, EP/R030073/1 (Tarapore). The authors acknowledge the IRIDIS High Performance Computing Facility and thank Arnold Benedict for initial work on the simulator. Publisher Copyright: © 2021 Owner/Author.

Keywords: damage recovery, evolutionary robotics, meta-evolution, quality-diversity algorithms, representational capacity

Learn more about Agents, Interactions and Complexity research