Quality-diversity in dissimilarity spaces

TL;DR

This paper introduces a general framework for quality-diversity algorithms based on the theory of magnitude, applied to dissimilarity spaces, and demonstrates a versatile version of Go-Explore with strong results.

Contribution

It extends quality-diversity algorithms to generic dissimilarity spaces using the theory of magnitude, enabling broader applicability and improved performance.

Findings

Demonstrates a general version of Go-Explore in dissimilarity spaces

Achieves promising performance on benchmark tasks

Provides a theoretical foundation for diversity optimization

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.