Subgoal-Based Temporal Abstraction in Monte-Carlo Tree Search

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.

Cite

Text

Gabor et al. "Subgoal-Based Temporal Abstraction in Monte-Carlo Tree Search." International Joint Conference on Artificial Intelligence, 2019. doi:10.24963/IJCAI.2019/772

Markdown

[Gabor et al. "Subgoal-Based Temporal Abstraction in Monte-Carlo Tree Search." International Joint Conference on Artificial Intelligence, 2019.](https://mlanthology.org/ijcai/2019/gabor2019ijcai-subgoal/) doi:10.24963/IJCAI.2019/772

BibTeX

@inproceedings{gabor2019ijcai-subgoal,
  title     = {{Subgoal-Based Temporal Abstraction in Monte-Carlo Tree Search}},
  author    = {Gabor, Thomas and Peter, Jan and Phan, Thomy and Meyer, Christian and Linnhoff-Popien, Claudia},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2019},
  pages     = {5562-5568},
  doi       = {10.24963/IJCAI.2019/772},
  url       = {https://mlanthology.org/ijcai/2019/gabor2019ijcai-subgoal/}
}