HSA-Net: Hierarchical and Structure-Aware Framework for Efficient and Scalable Molecular Language Modeling
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (2) · 980-987
DOI: 10.1609/aaai.v40i2.37067 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zihang Shao und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Molecular representation learning, a cornerstone for downstream tasks like molecular captioning and molecular property prediction, heavily relies on Graph Neural Networks (GNN). However, GNN suffers from the over-smoothing problem, where node-level features collapse in deep GNN layers. While existing feature projection methods with cross-attention have been introduced to mitigate this issue, they still perform poorly in deep features. This motivated our exploration of using Mamba as an alternative projector for its ability to handle complex sequences. However, we observe that while Mamba excels at preserving global topological information from deep layers, it neglects fine-grained details in shallow layers. The capabilities of Mamba and cross-attention exhibit a global-local trade-off. To resolve this critical global-local trade-off, we propose Hierarchical and Structure-Aware Network (HSA-Net), a novel framework with two modules that enables a hierarchical feature projection and fusion. Firstly, a Hierarchical Adaptive Projector (HAP) module is introduced to process features from different graph layers. It learns to dynamically switch between a cross-attention projector for shallow layers and a structure-aware Graph-Mamba projector for deep layers, producing high-quality, multi-level features. Secondly, to adaptively merge these multi-level features, we design a Source-Aware Fusion (SAF) module, which flexibly selects fusion experts based on the characteristics of the aggregation features, ensuring a precise and effective final representation fusion. Extensive experiments demonstrate that our HSA-Net framework quantitatively and qualitatively outperforms current state-of-the-art (SOTA) methods.
Zitieren
APA
Shao, Z., Lei, W., Wang, L., Ye, W., & Liu, L. (2026). HSA-Net: Hierarchical and Structure-Aware Framework for Efficient and Scalable Molecular Language Modeling. Proceedings of the AAAI Conference on Artificial Intelligence, 40(2), 980-987. https://doi.org/10.1609/aaai.v40i2.37067
Vancouver
Shao Z, Lei W, Wang L, Ye W, Liu L. HSA-Net: Hierarchical and Structure-Aware Framework for Efficient and Scalable Molecular Language Modeling. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(2):980-987. doi:10.1609/aaai.v40i2.37067
BibTeX
@article{Shao2026,
author = {Shao, Zihang and Lei, Wentao and Wang, Lei and Ye, Wen-Cai and Liu, Li},
title = {HSA-Net: Hierarchical and Structure-Aware Framework for Efficient and Scalable Molecular Language Modeling},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {2},
pages = {980--987},
doi = {10.1609/aaai.v40i2.37067}
}
EUVIMED