Beyond Single-Speed Reasoning: Coordinating Fast and Slow Dynamics for Efficient World Modeling
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (31) · 26215-26223
DOI: 10.1609/aaai.v40i31.39825 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Hongwei Wang und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Model-based reinforcement learning (MBRL) enables efficient decision-making by learning predictive world modelsof environment dynamics. Despite recent advances, existingmodels often struggle to reconcile accurate short-term transitions with coherent long-term planning, especially in partially observable or long-horizon settings. We argue that thislimitation often stems from modeling all transitions at a single temporal resolution, which makes it challenging to simultaneously capture fine-grained local dynamics and abstractglobal structures. To this end, we propose SF-RSSM (Slow-Fast Recurrent State-Space Model), a novel method that decouples short-term and long-term dynamics via a dualbranchdesign. The fast branch captures short-horizon transitions using residual prediction, while the slow branch models long-range dependencies with a GRU-based recurrent pathway.A distillation mechanism is developed to enable cooperationacross timescales, with the slow model providing soft targetsto guide the fast model. Additionally, a curiosity module encourages exploration by promoting learning in regions wherethe fast and slow branches exhibit divergent dynamics. Experiments on CARLA, DMControl and Atari benchmarks showthat SF-RSSM outperforms strong baselines in policy performance.
Zitieren
APA
Wang, H., Huang, Y., Chen, G., Wang, X., & Jin, Y. (2026). Beyond Single-Speed Reasoning: Coordinating Fast and Slow Dynamics for Efficient World Modeling. Proceedings of the AAAI Conference on Artificial Intelligence, 40(31), 26215-26223. https://doi.org/10.1609/aaai.v40i31.39825
Vancouver
Wang H, Huang Y, Chen G, Wang X, Jin Y. Beyond Single-Speed Reasoning: Coordinating Fast and Slow Dynamics for Efficient World Modeling. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(31):26215-26223. doi:10.1609/aaai.v40i31.39825
BibTeX
@article{Wang2026,
author = {Wang, Hongwei and Huang, Yangru and Chen, Guangyao and Wang, Xu and Jin, Yi},
title = {Beyond Single-Speed Reasoning: Coordinating Fast and Slow Dynamics for Efficient World Modeling},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {31},
pages = {26215--26223},
doi = {10.1609/aaai.v40i31.39825}
}
EUVIMED