SE(3)-equivariant Graph Neural Networks for Learning Glassy Liquids Representations - Laboratoire Interdisciplinaire des Sciences du Numérique Access content directly
Preprints, Working Papers, ... (Preprint) Year : 2022

SE(3)-equivariant Graph Neural Networks for Learning Glassy Liquids Representations

Abstract

Within the glassy liquids community, the use of Machine Learning (ML) to model particles' static structure in order to predict their future dynamics is currently a hot topic. The actual state of the art consists in Graph Neural Networks (GNNs) [BKGB + 20] which, beside having a great expressive power, are heavy models with numerous parameters and lack interpretability. Inspired by recent advances [TSK + 18], we build a GNN that learns a robust representation of the glass' static structure by constraining it to preserve the roto-translation (SE(3)) equivariance. We show that this constraint not only significantly improves the predictive power but also allows to reduce the number of parameters while improving the interpretability. Furthermore, we relate our learned equivariant features to well-known invariant expert features, which are easily expressible with a single layer of our network.
Fichier principal
Vignette du fichier
2211.03226v2.pdf (1.51 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03868206 , version 1 (23-11-2022)
hal-03868206 , version 2 (17-01-2024)

Licence

Attribution

Identifiers

Cite

Francesco Saverio Pezzicoli, Guillaume Charpiat, François P. Landes. SE(3)-equivariant Graph Neural Networks for Learning Glassy Liquids Representations. 2022. ⟨hal-03868206v2⟩
55 View
49 Download

Altmetric

Share

Gmail Facebook X LinkedIn More