Information Set MCTS for Hearts with Transposition Detection and Data-Driven Reward Estimation
Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · 08.11.2026 · 22 (1) · 222-229
DOI: 10.1609/aiide.v22i1.43580 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, online erschienen am 08.11.2026. Verfasst von Henrique Borges Manzke und 1 weiteren Autor:innen.
Abstract (Originalsprache)
This work presents an Information Set Monte Carlo Tree Search (MCTS) algorithm specifically tailored to the imperfect information card game of Hearts. The algorithm introduces a transposition detection mechanism for Hearts that groups information set nodes that, despite having different card distributions, share the same strategic context when card play order is disregarded, reducing computational redundancy and improving search efficiency. Additionally, a K-Nearest Neighbors (kNN) based reward function leverages historical data from both human and self-play games to estimate the likely final standing of a player given a score state, enabling more accurate value estimation in games where simulating until the final outcome is computationally infeasible. Experimental tests were conducted in a controlled environment to evaluate the impact of these contributions on playing strength. The results demonstrate that both proposed approaches significantly outperform their baseline versions.
Zitieren
APA
Manzke, H. B., & Tavares, A. R. (2026). Information Set MCTS for Hearts with Transposition Detection and Data-Driven Reward Estimation. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 22(1), 222-229. https://doi.org/10.1609/aiide.v22i1.43580
Vancouver
Manzke HB, Tavares AR. Information Set MCTS for Hearts with Transposition Detection and Data-Driven Reward Estimation. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment. 2026;22(1):222-229. doi:10.1609/aiide.v22i1.43580
BibTeX
@article{Manzke2026,
author = {Manzke, Henrique Borges and Tavares, Anderson Rocha},
title = {Information Set MCTS for Hearts with Transposition Detection and Data-Driven Reward Estimation},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment},
year = {2026},
volume = {22},
number = {1},
pages = {222--229},
doi = {10.1609/aiide.v22i1.43580}
}
EUVIMED