{"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/verifiable-reinforcement-learning-via-policy","title":"Verifiable Reinforcement Learning via Policy Extraction","arxiv_id":"1805.08328","date":"2018-05-22","proceeding":"NeurIPS 2018 12","authors":["Osbert Bastani","Yewen Pu","Armando Solar-Lezama"],"abstract":"While deep reinforcement learning has successfully solved many challenging\ncontrol tasks, its real-world applicability has been limited by the inability\nto ensure the safety of learned policies. We propose an approach to verifiable\nreinforcement learning by training decision tree policies, which can represent\ncomplex policies (since they are nonparametric), yet can be efficiently\nverified using existing techniques (since they are highly structured). The\nchallenge is that decision tree policies are difficult to train. We propose\nVIPER, an algorithm that combines ideas from model compression and imitation\nlearning to learn decision tree policies guided by a DNN policy (called the\noracle) and its Q-function, and show that it substantially outperforms two\nbaselines. We use VIPER to (i) learn a provably robust decision tree policy for\na variant of Atari Pong with a symbolic state space, (ii) learn a decision tree\npolicy for a toy game based on Pong that provably never loses, and (iii) learn\na provably stable decision tree policy for cart-pole. In each case, the\ndecision tree policy achieves performance equal to that of the original DNN\npolicy.","url_abs":"http://arxiv.org/abs/1805.08328v2","url_pdf":"http://arxiv.org/pdf/1805.08328v2.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":"verifiable-reinforcement-learning-via-policy","repo_url":"https://github.com/quantumiracle/Cascading-Decision-Tree","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"verifiable-reinforcement-learning-via-policy","repo_url":"https://github.com/safe-rl-team/viper-verifiable-rl-impl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"imitation-learning","task_name":"Imitation Learning"},{"task_slug":"model-compression","task_name":"Model Compression"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.08328","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}