{"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/meta-reinforcement-learning-of-structured","title":"Meta-Reinforcement Learning of Structured Exploration Strategies","arxiv_id":"1802.07245","date":"2018-02-20","proceeding":"NeurIPS 2018 12","authors":["Abhishek Gupta","Russell Mendonca","Yuxuan Liu","Pieter Abbeel","Sergey Levine"],"abstract":"Exploration is a fundamental challenge in reinforcement learning (RL). Many\nof the current exploration methods for deep RL use task-agnostic objectives,\nsuch as information gain or bonuses based on state visitation. However, many\npractical applications of RL involve learning more than a single task, and\nprior tasks can be used to inform how exploration should be performed in new\ntasks. In this work, we explore how prior tasks can inform an agent about how\nto explore effectively in new situations. We introduce a novel gradient-based\nfast adaptation algorithm -- model agnostic exploration with structured noise\n(MAESN) -- to learn exploration strategies from prior experience. The prior\nexperience is used both to initialize a policy and to acquire a latent\nexploration space that can inject structured stochasticity into a policy,\nproducing exploration strategies that are informed by prior knowledge and are\nmore effective than random action-space noise. We show that MAESN is more\neffective at learning exploration strategies when compared to prior meta-RL\nmethods, RL without learned exploration strategies, and task-agnostic\nexploration methods. We evaluate our method on a variety of simulated tasks:\nlocomotion with a wheeled robot, locomotion with a quadrupedal walker, and\nobject manipulation.","url_abs":"http://arxiv.org/abs/1802.07245v1","url_pdf":"http://arxiv.org/pdf/1802.07245v1.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":"meta-reinforcement-learning-of-structured","repo_url":"https://github.com/russellmendonca/maesn_suite","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"meta-reinforcement-learning-of-structured","repo_url":"https://github.com/llan-ml/tesp","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"meta-reinforcement-learning","task_name":"Meta Reinforcement Learning"},{"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=1802.07245","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}