EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Multi-Level Reinforcement Learning Reasoning Model Based on Adaptive Hierarchical Feature Fusion

Siyu Zhu, Cheng He, Yanlan Wang, Qitao Tai, Lin Miao, Liang Wang, Xiulei Liu, Yonghao Yang, Jiahao Zhang, Weijian Guo

Symmetry · 28.06.2026 · 18 (7) · 1097

DOI: 10.3390/sym18071097 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Symmetry, online erschienen am 28.06.2026. Verfasst von Siyu Zhu und 9 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key Laboratory of Complex Aviation System Simulation Open Fund, Young Elite Scientists Sponsorship Program of the Beijing High Innovation Plan.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Current knowledge graph reasoning methods actually use the same state representations and a fixed decision method to complete the total reasoning process. They are not able to truly use the different levels of attribute information. To overcome this problem, we propose a multi-level reinforcement learning reasoning model based on adaptive hierarchical feature fusion. Our model divides attribute information into coarse-grained attributes and fine-grained attributes. In this way, our model can better balance overall semantics and details, thereby holding connections between entities and relationships. We design an adaptive hierarchical feature fusion mechanism. Our model dynamically determines whether it pays more attention to attribute information or relationship according to the needs of the current task, rather than constantly using the same weights. This enables our model to have better performance in different reasoning situations. We improve the decision process of the model, making the model know which relationships to focus on and which are less relevant to skip, thereby reducing ineffective search. According to the results of the experiment, our method is superior to the existing reinforcement learning reasoning methods, including MINERVA, AdaPath, CURL, and HMR. Our model improves MRR, Hits@1, Hits@3, and Hits@10 indicators compared with these baseline models. This shows that our method can improve the accuracy of inference and verify the effectiveness of our model in complex knowledge graph inference tasks.

Förderung

  • National Key Laboratory of Complex Aviation System Simulation Open Fund
  • Young Elite Scientists Sponsorship Program of the Beijing High Innovation Plan

Zitieren

APA

Zhu, S., He, C., Wang, Y., Tai, Q., Miao, L., Wang, L., Liu, X., Yang, Y., Zhang, J., & Guo, W. (2026). A Multi-Level Reinforcement Learning Reasoning Model Based on Adaptive Hierarchical Feature Fusion. Symmetry, 18(7), 1097. https://doi.org/10.3390/sym18071097

Vancouver

Zhu S, He C, Wang Y, Tai Q, Miao L, Wang L, et al. A Multi-Level Reinforcement Learning Reasoning Model Based on Adaptive Hierarchical Feature Fusion. Symmetry. 2026;18(7):1097. doi:10.3390/sym18071097

BibTeX
@article{Zhu2026,
  author = {Zhu, Siyu and He, Cheng and Wang, Yanlan and Tai, Qitao and Miao, Lin and Wang, Liang and Liu, Xiulei and Yang, Yonghao and Zhang, Jiahao and Guo, Weijian},
  title = {A Multi-Level Reinforcement Learning Reasoning Model Based on Adaptive Hierarchical Feature Fusion},
  journal = {Symmetry},
  year = {2026},
  volume = {18},
  number = {7},
  pages = {1097},
  doi = {10.3390/sym18071097}
}
Kostenlos starten →