{"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/switch-based-active-deep-dyna-q-efficient","title":"Switch-based Active Deep Dyna-Q: Efficient Adaptive Planning for Task-Completion Dialogue Policy Learning","arxiv_id":"1811.07550","date":"2018-11-19","proceeding":null,"authors":["Yuexin Wu","Xiujun Li","Jingjing Liu","Jianfeng Gao","Yiming Yang"],"abstract":"Training task-completion dialogue agents with reinforcement learning usually\nrequires a large number of real user experiences. The Dyna-Q algorithm extends\nQ-learning by integrating a world model, and thus can effectively boost\ntraining efficiency using simulated experiences generated by the world model.\nThe effectiveness of Dyna-Q, however, depends on the quality of the world model\n- or implicitly, the pre-specified ratio of real vs. simulated experiences used\nfor Q-learning. To this end, we extend the recently proposed Deep Dyna-Q (DDQ)\nframework by integrating a switcher that automatically determines whether to\nuse a real or simulated experience for Q-learning. Furthermore, we explore the\nuse of active learning for improving sample efficiency, by encouraging the\nworld model to generate simulated experiences in the state-action space where\nthe agent has not (fully) explored. Our results show that by combining switcher\nand active learning, the new framework named as Switch-based Active Deep Dyna-Q\n(Switch-DDQ), leads to significant improvement over DDQ and Q-learning\nbaselines in both simulation and human evaluations.","url_abs":"http://arxiv.org/abs/1811.07550v1","url_pdf":"http://arxiv.org/pdf/1811.07550v1.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":"switch-based-active-deep-dyna-q-efficient","repo_url":"https://github.com/CrickWu/Swtich-DDQ","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"active-learning","task_name":"Active Learning"},{"task_slug":"q-learning","task_name":"Q-Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"task-completion-dialogue-policy-learning","task_name":"Task-Completion Dialogue Policy Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.07550","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}