Multi-Task Multi-Behavior MAP-Elites

Abstract

We propose Multi-Task Multi-Behavior MAP-Elites, a variant of MAP-Elites that finds a large number of high-quality solutions for a large set of tasks (optimization problems from a given family). It combines the original MAP-Elites for the search for diversity and Multi-Task MAP-Elites for leveraging similarity between tasks. It performs better than three baselines on a humanoid fault-recovery set of tasks, solving more tasks and finding twice as many solutions per solved task.

arXiv commentAccepted as Poster for GECCO 2023

References in corpus (4)

• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
• Illuminating search spaces by mapping elites
• Quality Diversity for Multi-task Optimization
• First do not fall: learning to exploit a wall with a damaged humanoid robot

Cited by in corpus (1)

• Parametric-Task MAP-Elites