{"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/transfer-of-deep-reactive-policies-for-mdp","title":"Transfer of Deep Reactive Policies for MDP Planning","arxiv_id":"1810.11488","date":"2018-10-26","proceeding":"NeurIPS 2018 12","authors":["Aniket Bajpai","Sankalp Garg","Mausam"],"abstract":"Domain-independent probabilistic planners input an MDP description in a\nfactored representation language such as PPDDL or RDDL, and exploit the\nspecifics of the representation for faster planning. Traditional algorithms\noperate on each problem instance independently, and good methods for\ntransferring experience from policies of other instances of a domain to a new\ninstance do not exist. Recently, researchers have begun exploring the use of\ndeep reactive policies, trained via deep reinforcement learning (RL), for MDP\nplanning domains. One advantage of deep reactive policies is that they are more\namenable to transfer learning.\n  In this paper, we present the first domain-independent transfer algorithm for\nMDP planning domains expressed in an RDDL representation. Our architecture\nexploits the symbolic state configuration and transition function of the domain\n(available via RDDL) to learn a shared embedding space for states and\nstate-action pairs for all problem instances of a domain. We then learn an RL\nagent in the embedding space, making a near zero-shot transfer possible, i.e.,\nwithout much training on the new instance, and without using the domain\nsimulator at all. Experiments on three different benchmark domains underscore\nthe value of our transfer algorithm. Compared against planning from scratch,\nand a state-of-the-art RL transfer algorithm, our transfer solution has\nsignificantly superior learning curves.","url_abs":"http://arxiv.org/abs/1810.11488v1","url_pdf":"http://arxiv.org/pdf/1810.11488v1.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":"transfer-of-deep-reactive-policies-for-mdp","repo_url":"https://github.com/dair-iitd/torpido","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1810.11488","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}