References & Citations
Computer Science > Neural and Evolutionary Computing
Title: Evolutionary Reinforcement Learning via Cooperative Coevolution
(Submitted on 23 Apr 2024 (v1), last revised 29 Apr 2024 (this version, v2))
Abstract: Recently, evolutionary reinforcement learning has obtained much attention in various domains. Maintaining a population of actors, evolutionary reinforcement learning utilises the collected experiences to improve the behaviour policy through efficient exploration. However, the poor scalability of genetic operators limits the efficiency of optimising high-dimensional neural networks. To address this issue, this paper proposes a novel cooperative coevolutionary reinforcement learning (CoERL) algorithm. Inspired by cooperative coevolution, CoERL periodically and adaptively decomposes the policy optimisation problem into multiple subproblems and evolves a population of neural networks for each of the subproblems. Instead of using genetic operators, CoERL directly searches for partial gradients to update the policy. Updating policy with partial gradients maintains consistency between the behaviour spaces of parents and offspring across generations. The experiences collected by the population are then used to improve the entire policy, which enhances the sampling efficiency. Experiments on six benchmark locomotion tasks demonstrate that CoERL outperforms seven state-of-the-art algorithms and baselines. Ablation study verifies the unique contribution of CoERL's core ingredients.
Submission history
From: Chengpeng Hu [view email][v1] Tue, 23 Apr 2024 05:56:35 GMT (7390kb,D)
[v2] Mon, 29 Apr 2024 13:52:35 GMT (7390kb,D)
Link back to: arXiv, form interface, contact.