Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
Abstract
Informative uncertainty quantification (UQ) is essential for the practical deployment of machine-learned interatomic potentials (MLIPs), as it provides insight into their reliability across chemical space. Although numerous post-hoc uncertainty estimation methods for MLIPs have recently been proposed, most have been trained in low-data regimes, restricting their applicability to narrow domains of interest. Inspired by the emergence of foundational MLIPs, we propose pre-training uncertainty estimators on large, diverse datasets to enable broader applicability. We pre-train several UQ models on large materials science datasets to quantify the uncertainty of different MLIPs, and evaluate their performance on both in-distribution and out-of-domain (OOD) adsorption-energy predictions. We find that pre-trained uncertainty estimators provide viable zero-shot uncertainty estimates across multiple datasets, while fine-tuning consistently yields positive correlations between predicted uncertainty and prediction error, even on the most challenging datasets. These results lay the groundwork for foundational UQ models for MLIPs, ultimately facilitating their deployment in practical settings by providing users with a priori information about model reliability and applicability.
Related
- Machine Learning Interatomic Potentials for Energy Materials: Architectures, Training Strategies, and Applications
- AdaptNTK: Adaptive Uncertainty Quantification and Active Learning for Neural Network Potentials
- Correcting the Liquid Densities of Universal Machine-Learning Interatomic Potentials Using Local-Number-Density Repulsion
- Foundation interatomic potentials learn the practical chemistry they were never taught
- pdb2reaction: End-to-End Reaction-Path Elucidation from PDB Structures Using Machine-Learning Interatomic Potentials
BibTeX
@misc{schmid2026towards,
title = {{Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training}},
author = {Stefan P. Schmid and Tobias Setzer and Maximilian Scheurer},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.41801},
eprint = {2610.41801},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41801},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
AU - Stefan P. Schmid
AU - Tobias Setzer
AU - Maximilian Scheurer
PY - 2026
DA - 2026/10/30
PB - arXiv
DO - 10.48550/arxiv.2610.41801
UR - https://arxiv.org/abs/2610.41801
AB - Informative uncertainty quantification (UQ) is essential for the practical deployment of machine-learned interatomic potentials (MLIPs), as it provides insight into their reliability across chemical space. Although numerous post-hoc uncertainty estimation methods for MLIPs have recently been proposed, most have been trained in low-data regimes, restricting their applicability to narrow domains of interest. Inspired by the emergence of foundational MLIPs, we propose pre-training uncertainty estimators on large, diverse datasets to enable broader applicability. We pre-train several UQ models on large materials science datasets to quantify the uncertainty of different MLIPs, and evaluate their performance on both in-distribution and out-of-domain (OOD) adsorption-energy predictions. We find that pre-trained uncertainty estimators provide viable zero-shot uncertainty estimates across multiple datasets, while fine-tuning consistently yields positive correlations between predicted uncertainty and prediction error, even on the most challenging datasets. These results lay the groundwork for foundational UQ models for MLIPs, ultimately facilitating their deployment in practical settings by providing users with a priori information about model reliability and applicability.
ER -