{"type":"ScholarlyArticle","title":"ASAP-UCT: Abstraction of State-Action Pairs in UCT","authors":[{"name":"Ankit Anand","slug":"ankit-anand","given":"Ankit","family":"Anand","affiliation":""},{"name":"Aditya Grover","slug":"aditya-grover","given":"Aditya","family":"Grover","affiliation":""},{"name":"Mausam","slug":"mausam","given":"","family":"Mausam","affiliation":""},{"name":"Parag Singla","slug":"parag-singla","given":"Parag","family":"Singla","affiliation":""}],"abstract":"Monte-Carlo Tree Search (MCTS) algorithms such as UCT are an attractive online framework for solving planning under uncertainty problems modeled as a Markov Decision Process. However, MCTS search trees are constructed in flat state and action spaces, which can lead to poor policies for large problems. In a separate research thread, domain abstraction techniques compute symmetries to reduce the original MDP. This can lead to significant savings in computation, but these have been predominantly implemented for offline planning. This paper makes two contributions. First, we define the ASAP (Abstraction of State-Action Pairs) framework, which extends and unifies past work on domain abstractions by holistically aggregating both states and state-action pairs — ASAP uncovers a much larger number of symmetries in a given domain. Second, we propose ASAP-UCT, which implements ASAP-style abstractions within a UCT framework combining strengths of online planning with domain abstractions. Experimental evaluation on several benchmark domains shows up to 26% improvement in the quality of policies obtained over existing algorithms.","doi":null,"url":"https://ijcai.org/proceedings/2015/216","pdf":"https://ijcai.org/proceedings/2015/0216.pdf","markdown":"https://ijcai.org/proceedings/2015/216.md","bibtex":"https://ijcai.org/proceedings/2015/bibtex/216","volume":{"slug":"2015","title":"Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2015","url":"https://ijcai.org/proceedings/2015/","year":2015,"location":"Buenos Aires, Argentina","dates":"25–31 July 2015","editors":"Qiang Yang and Michael Wooldridge","isbn":"978-1-57735-738-4","publisher":"International Joint Conferences on Artificial Intelligence","doi_prefix":""},"section":"Planning","track":"Main Track","number":216,"pages":{"first":1509,"last":1514},"keywords":[],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}