Official Review · ICLR 2024

This paper proposes the cooperative coevolution for QD methods on reinforcement learning problems. The major contribution is claimed to be the first time use of cooperative coevolution, and this can help improve the sample efficiency. The experiment setup is considerate, and the results are promising. The main idea of using cooperative coevolution into reinforcement learning is easy to understand for addressing large-scale search space. This work is not the first time of applying cooperative coevolution to reinforcement learning, even restricted in the quality-diversity like evolutionary algorithms, see the following reference: [1] Evolutionary reinforcement learning via cooperative coevolutionary negatively correlated search. So the related claims should be revised. Given the above work, the contribution of this work may be limited. What is the relationship between large-scale search space and the sample efficiency? Why address the large-scale space with cooperative coevolution can help improve the sample efficiency of QD?