{"type":"ScholarlyArticle","title":"Don’t Do What Doesn’t Matter: Intrinsic Motivation with Action Usefulness","authors":[{"name":"Mathieu Seurin","slug":"mathieu-seurin","given":"Mathieu","family":"Seurin","affiliation":""},{"name":"Florian Strub","slug":"florian-strub","given":"Florian","family":"Strub","affiliation":""},{"name":"Philippe Preux","slug":"philippe-preux","given":"Philippe","family":"Preux","affiliation":""},{"name":"Olivier Pietquin","slug":"olivier-pietquin","given":"Olivier","family":"Pietquin","affiliation":""}],"abstract":"Sparse rewards are double-edged training signals in reinforcement learning: \r\neasy to design but hard to optimize. Intrinsic motivation guidances have thus been developed toward alleviating the resulting exploration problem. They usually incentivize agents to look for new states through novelty signals. Yet, such methods encourage exhaustive exploration of the state space rather than focusing on the environment's salient interaction opportunities. We propose a new exploration method, called Don't Do What Doesn't Matter (DoWhaM), shifting the emphasis from state novelty to state with relevant actions. While most actions consistently change the state when used, e.g. moving the agent, some actions are only effective in specific states, e.g., opening a door, grabbing an object. DoWhaM detects and rewards actions that seldom affect the environment.\r\nWe evaluate DoWhaM on the procedurally-generated environment MiniGrid against state-of-the-art methods. Experiments consistently show that DoWhaM greatly reduces sample complexity, installing the new state-of-the-art in MiniGrid.","doi":"10.24963/ijcai.2021/406","url":"https://ijcai.org/proceedings/2021/406","pdf":"https://ijcai.org/proceedings/2021/0406.pdf","markdown":"https://ijcai.org/proceedings/2021/406.md","bibtex":"https://ijcai.org/proceedings/2021/bibtex/406","volume":{"slug":"2021","title":"Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2021","url":"https://ijcai.org/proceedings/2021/","year":2021,"location":"Montreal","dates":"19-27 August 2021","editors":"Prof. Zhi-Hua Zhou","isbn":"978-0-9992411-9-6","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2021"},"section":"Machine Learning","track":"Main Track","number":406,"pages":{"first":2950,"last":2956},"keywords":["Machine Learning: Deep Reinforcement Learning","Machine Learning: Reinforcement Learning"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}