EUVIMEDEUVIMED entdecken
Crossref

Magnitude-Modulated Equivariant Adapter for Parameter-Efficient Fine-Tuning of Equivariant Graph Neural Networks

Dian Jin, Yancheng Yuan, Xiaoming Tao

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (1) · 498-506

DOI: 10.1609/aaai.v40i1.37013 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Dian Jin und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Pretrained equivariant graph neural networks based on spherical harmonics offer efficient and accurate alternatives to computationally expensive ab-initio methods, yet adapting them to new tasks and chemical environments still requires fine-tuning. Conventional parameter-efficient fine-tuning (PEFT) techniques, such as Adapters and LoRA, typically break symmetry, making them incompatible with those equivariant architectures. ELoRA, recently proposed, is the first equivariant PEFT method. It achieves improved parameter efficiency and performance on many benchmarks. However, the relatively high degrees of freedom it retains within each tensor order can still perturb pretrained feature distributions and ultimately degrade performance. To address this, we present Magnitude-Modulated Equivariant Adapter (MMEA), a novel equivariant fine-tuning method which employs lightweight scalar gating to modulate feature magnitudes on a per-order and per-multiplicity basis. We demonstrate that MMEA preserves strict equivariance and, across multiple benchmarks, consistently improves energy and force predictions to state-of-the-art levels while training fewer parameters than competing approaches. These results suggest that, in many practical scenarios, modulating channel magnitudes is sufficient to adapt equivariant models to new chemical environments without breaking symmetry, pointing toward a new paradigm for equivariant PEFT design.

Zitieren

APA

Jin, D., Yuan, Y., & Tao, X. (2026). Magnitude-Modulated Equivariant Adapter for Parameter-Efficient Fine-Tuning of Equivariant Graph Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 40(1), 498-506. https://doi.org/10.1609/aaai.v40i1.37013

Vancouver

Jin D, Yuan Y, Tao X. Magnitude-Modulated Equivariant Adapter for Parameter-Efficient Fine-Tuning of Equivariant Graph Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(1):498-506. doi:10.1609/aaai.v40i1.37013

BibTeX
@article{Jin2026,
  author = {Jin, Dian and Yuan, Yancheng and Tao, Xiaoming},
  title = {Magnitude-Modulated Equivariant Adapter for Parameter-Efficient Fine-Tuning of Equivariant Graph Neural Networks},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {1},
  pages = {498--506},
  doi = {10.1609/aaai.v40i1.37013}
}
Kostenlos starten →