We propose an approach to general subgoal-based temporal abstraction in MCTS. Our approach approximates a set of available macro-actions locally for each state only requiring a generative model and a subgoal predicate. For that, we modify the expansion step of MCTS to automatically discover and optimize macro-actions that lead to subgoals. We empirically evaluate the effectiveness, computational efficiency and robustness of our approach w.r.t. different parameter settings in two benchmark domains and compare the results to standard MCTS without temporal abstraction.
@inproceedings{ gaborIJCAI19,
author = "Thomas Gabor and Jan Peter and Thomy Phan and Christian Meyer and Claudia Linnhoff-Popien",
title = "Subgoal-Based Temporal Abstraction in Monte-Carlo Tree Search",
year = "2019",
abstract = "We propose an approach to general subgoal-based temporal abstraction in MCTS. Our approach approximates a set of available macro-actions locally for each state only requiring a generative model and a subgoal predicate. For that, we modify the expansion step of MCTS to automatically discover and optimize macro-actions that lead to subgoals. We empirically evaluate the effectiveness, computational efficiency and robustness of our approach w.r.t. different parameter settings in two benchmark domains and compare the results to standard MCTS without temporal abstraction.",
url = "https://www.ijcai.org/proceedings/2019/772",
eprint = "https://thomyphan.github.io/files/2019-ijcai-2.pdf",
publisher = "International Joint Conferences on Artificial Intelligence Organization",
booktitle = "Proceedings of the 28th International Joint Conference on Artificial Intelligence",
pages = "5562--5568",
doi = "https://doi.org/10.24963/ijcai.2019/772"
}
Related Articles
Relevant Research Areas