Black-Box Uncertainty Estimation for Deep Learning Models in Atomistic Simulations

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fonea, Idan, Peles, Amir, Niv, Sivan, Gordon, Goren, Natan, Amir
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912721051582464
author Fonea, Idan
Peles, Amir
Niv, Sivan
Gordon, Goren
Natan, Amir
author_facet Fonea, Idan
Peles, Amir
Niv, Sivan
Gordon, Goren
Natan, Amir
contents We analyze an ensemble-based approach for uncertainty quantification (UQ) in atomistic neural networks. This method generates an epistemic uncertainty signal without requiring changes to the underlying multi-headed regression neural network architecture, making it suitable for sealed or black-box models. We apply this method to molecular systems, specifically sodium (Na) and aluminum (Al), under various temperature conditions. By scaling the uncertainty signal, we account for heteroscedasticity in the data. We demonstrate the robustness of the scaled UQ signal for detecting out-of-distribution (OOD) behavior in several scenarios. This UQ signal also correlates with model convergence during training, providing an additional tool for optimizing the training process.
format Preprint
id arxiv_https___arxiv_org_abs_2511_16439
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Black-Box Uncertainty Estimation for Deep Learning Models in Atomistic Simulations
Fonea, Idan
Peles, Amir
Niv, Sivan
Gordon, Goren
Natan, Amir
Chemical Physics
We analyze an ensemble-based approach for uncertainty quantification (UQ) in atomistic neural networks. This method generates an epistemic uncertainty signal without requiring changes to the underlying multi-headed regression neural network architecture, making it suitable for sealed or black-box models. We apply this method to molecular systems, specifically sodium (Na) and aluminum (Al), under various temperature conditions. By scaling the uncertainty signal, we account for heteroscedasticity in the data. We demonstrate the robustness of the scaled UQ signal for detecting out-of-distribution (OOD) behavior in several scenarios. This UQ signal also correlates with model convergence during training, providing an additional tool for optimizing the training process.
title Black-Box Uncertainty Estimation for Deep Learning Models in Atomistic Simulations
topic Chemical Physics
url https://arxiv.org/abs/2511.16439