fig7

Advances in graph neural networks for alloy design and properties predictions: a review

Figure 7. (A) ALIGNN architecture, which augments the atom-level graph with a line graph so that bond and bond-angle information is captured through coupled message-passing updates of atoms, bonds, and angles; (B) Efficiency-optimized graph neural model developed by Dong, featuring radial-basis edge encodings and adaptive interaction blocks that accelerate training while preserving accuracy. (A) reproduced from Ref.[48], copyright © 2021. (B) reproduced with permission from Ref.[49], copyright © 2023 American Chemical Society. ALIGNN: Atomistic line graph neural network.

Journal of Materials Informatics
ISSN 2770-372X (Online)
Follow Us

Portico

All published articles are preserved here permanently:

https://www.portico.org/publishers/oae/

Portico

All published articles are preserved here permanently:

https://www.portico.org/publishers/oae/