{"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/safe-model-based-reinforcement-learning-with","title":"Safe Model-based Reinforcement Learning with Stability Guarantees","arxiv_id":"1705.08551","date":"2017-05-23","proceeding":"NeurIPS 2017 12","authors":["Felix Berkenkamp","Matteo Turchetta","Angela P. Schoellig","Andreas Krause"],"abstract":"Reinforcement learning is a powerful paradigm for learning optimal policies\nfrom experimental data. However, to find optimal policies, most reinforcement\nlearning algorithms explore all possible actions, which may be harmful for\nreal-world systems. As a consequence, learning algorithms are rarely applied on\nsafety-critical systems in the real world. In this paper, we present a learning\nalgorithm that explicitly considers safety, defined in terms of stability\nguarantees. Specifically, we extend control-theoretic results on Lyapunov\nstability verification and show how to use statistical models of the dynamics\nto obtain high-performance control policies with provable stability\ncertificates. Moreover, under additional regularity assumptions in terms of a\nGaussian process prior, we prove that one can effectively and safely collect\ndata in order to learn about the dynamics and thus both improve control\nperformance and expand the safe region of the state space. In our experiments,\nwe show how the resulting algorithm can safely optimize a neural network policy\non a simulated inverted pendulum, without the pendulum ever falling down.","url_abs":"http://arxiv.org/abs/1705.08551v3","url_pdf":"http://arxiv.org/pdf/1705.08551v3.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":"safe-model-based-reinforcement-learning-with","repo_url":"https://github.com/befelix/safe_learning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"model-based-reinforcement-learning","task_name":"Model-based Reinforcement Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"model","task_name":"model"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1705.08551","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}