{"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/batch-policy-learning-under-constraints","title":"Batch Policy Learning under Constraints","arxiv_id":"1903.08738","date":"2019-03-20","proceeding":null,"authors":["Hoang M. Le","Cameron Voloshin","Yisong Yue"],"abstract":"When learning policies for real-world domains, two important questions arise:\n(i) how to efficiently use pre-collected off-policy, non-optimal behavior data;\nand (ii) how to mediate among different competing objectives and constraints.\nWe thus study the problem of batch policy learning under multiple constraints,\nand offer a systematic solution. We first propose a flexible meta-algorithm\nthat admits any batch reinforcement learning and online learning procedure as\nsubroutines. We then present a specific algorithmic instantiation and provide\nperformance guarantees for the main objective and all constraints. To certify\nconstraint satisfaction, we propose a new and simple method for off-policy\npolicy evaluation (OPE) and derive PAC-style bounds. Our algorithm achieves\nstrong empirical results in different domains, including in a challenging\nproblem of simulated car driving subject to multiple constraints such as lane\nkeeping and smooth driving. We also show experimentally that our OPE method\noutperforms other popular OPE techniques on a standalone basis, especially in a\nhigh-dimensional setting.","url_abs":"http://arxiv.org/abs/1903.08738v1","url_pdf":"http://arxiv.org/pdf/1903.08738v1.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":[{"paper_slug":"batch-policy-learning-under-constraints","repo_url":"https://github.com/gwthomas/gtml","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"batch-policy-learning-under-constraints","repo_url":"https://github.com/clvoloshin/constrained_batch_policy_learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.08738","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}