Differentiable Quality Diversity

TL;DR

This paper introduces the differentiable quality diversity (DQD) problem and presents MEGA, a gradient-based algorithm that significantly improves exploration efficiency in quality diversity tasks when gradients are available.

Contribution

The paper formulates the DQD problem and proposes MEGA, a novel gradient-based algorithm that outperforms existing methods in quality diversity optimization.

Findings

MEGA outperforms state-of-the-art QD algorithms in benchmark domains.

Gradient information significantly enhances exploration efficiency.

DQD enables more effective optimization in applications like StyleGAN latent space search.

Abstract

Quality diversity (QD) is a growing branch of stochastic optimization research that studies the problem of generating an archive of solutions that maximize a given objective function but are also diverse with respect to a set of specified measure functions. However, even when these functions are differentiable, QD algorithms treat them as "black boxes", ignoring gradient information. We present the differentiable quality diversity (DQD) problem, a special case of QD, where both the objective and measure functions are first order differentiable. We then present MAP-Elites via a Gradient Arborescence (MEGA), a DQD algorithm that leverages gradient information to efficiently explore the joint range of the objective and measure functions. Results in two QD benchmark domains and in searching the latent space of a StyleGAN show that MEGA significantly outperforms state-of-the-art QD algorithms, highlighting DQD's promise for efficient quality diversity optimization when gradient information is available. Source code is available at https://github.com/icaros-usc/dqd.