{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/off-policy-actor-critic-with-shared-1","title":"Off-Policy Actor-Critic with Shared Experience Replay","arxiv_id":"1909.11583","date":"2019-09-25","proceeding":"ICML 2020 1","authors":["Simon Schmitt","Matteo Hessel","Karen Simonyan"],"abstract":"We investigate the combination of actor-critic reinforcement learning algorithms with uniform large-scale experience replay and propose solutions for two challenges: (a) efficient actor-critic learning with experience replay (b) stability of off-policy learning where agents learn from other agents behaviour. We employ those insights to accelerate hyper-parameter sweeps in which all participating agents run concurrently and share their experience via a common replay module. To this end we analyze the bias-variance tradeoffs in V-trace, a form of importance sampling for actor-critic methods. Based on our analysis, we then argue for mixing experience sampled from replay with on-policy experience, and propose a new trust region scheme that scales effectively to data distributions where V-trace becomes unstable. We provide extensive empirical validation of the proposed solution. We further show the benefits of this setup by demonstrating state-of-the-art data efficiency on Atari among agents trained up until 200M environment frames.","url_abs":"https://arxiv.org/abs/1909.11583v2","url_pdf":"https://arxiv.org/pdf/1909.11583v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"atari-games","task_name":"Atari Games"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"}],"methods":[{"method_slug":"experience-replay","method_name":"Experience Replay"},{"method_slug":"v-trace","method_name":"V-trace"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/atari-games-on-atari-games","task":"Atari Games","dataset":"Atari games","model":"LASER","rank_in_archive_order":8,"of":12,"metrics":{"Mean Human Normalized Score":"1741.36%"},"uses_additional_data":false},{"leaderboard":"/sota/atari-games-on-atari-57","task":"Atari Games","dataset":"Atari-57","model":"LASER","rank_in_archive_order":7,"of":11,"metrics":{"Human World Record Breakthrough":"7","Mean Human Normalized Score":"1741.36%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1909.11583","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}