EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Multimodal Representation Learning Framework for Molecular Graph and NMR Spectrum Alignment

Xiao Li iD, Xun Wang, Zhong-Ming Liu, Jin-Biao Liu iD, Xin Huang iD

Entropy · 07.05.2026 · 28 (5) · 532

DOI: 10.3390/e28050532 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Entropy, online erschienen am 07.05.2026. Verfasst von Xiao Li und 4 weiteren Autor:innen. Beteiligte Einrichtung: School of Artificial Intelligence, Jiangxi Normal University, Nanchang 330022, China. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Jiangxi Province.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate matching between molecular structures and NMR spectra is an important task in automated structure elucidation. However, existing methods still face difficulties in jointly modeling multi-scale molecular topology and effectively exploiting the complementary information provided by paired 1H and 13C NMR spectra. To address these limitations, we propose SpecMol-MatchNet, a multimodal matching framework that integrates a hybrid molecular graph encoder, branch-specific spectral feature learning, and residual multimodal fusion. In the molecular branch, attention-based graph interaction is combined with multi-scale neighborhood aggregation to capture structural cues at different receptive fields. In the spectral branch, branch-specific attention enhancement and joint gating are introduced to better exploit the complementary characteristics of paired 1H and 13C spectra. The resulting molecular and spectral representations are integrated through a residual fusion module for final matching prediction. Experimental results on benchmark datasets demonstrate that SpecMol-MatchNet achieves consistently better overall performance than representative baseline methods.

Schlagwörter

Stichwörter der Autor:innen

cross-modal fusiongraph neural networksinformation bottleneck theorymultimodal representation learning

Autor:innen und Institutionen

  • School of Artificial Intelligence, Jiangxi Normal University, Nanchang 330022, China.
  • School of Economics and Management, Jiangxi Normal University, Nanchang 330022, China.
  • Jiangxi Provincial Key Laboratory of Functional Molecular Materials Chemistry, Jiangxi University of Science and Technology, Ganzhou 341000, China.
  • Jiangxi Provincial Engineering Research Center of Blockchain Data Security and Governance, Nanchang 330031, China.

ORCID-Profile: Xiao Li, Jin-Biao Liu, Xin Huang

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Jiangxi Province

Zitieren

APA

Li, X., Wang, X., Liu, Z. M., Liu, J. B., & Huang, X. (2026). A Multimodal Representation Learning Framework for Molecular Graph and NMR Spectrum Alignment. Entropy, 28(5), 532. https://doi.org/10.3390/e28050532

Vancouver

Li X, Wang X, Liu ZM, Liu JB, Huang X. A Multimodal Representation Learning Framework for Molecular Graph and NMR Spectrum Alignment. Entropy. 2026;28(5):532. doi:10.3390/e28050532

BibTeX
@article{Li2026,
  author = {Li, Xiao and Wang, Xun and Liu, Zhong-Ming and Liu, Jin-Biao and Huang, Xin},
  title = {A Multimodal Representation Learning Framework for Molecular Graph and NMR Spectrum Alignment},
  journal = {Entropy},
  year = {2026},
  volume = {28},
  number = {5},
  pages = {532},
  doi = {10.3390/e28050532}
}
Kostenlos starten →