Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On?
Abstract
Although machine learning potentials (MLPs) generalize across configurational and chemical compound spaces, their accuracy is commonly believed to be limited by their training methods, with improvements beyond this limit achieved so far only by refitting to experimental observables. In this Perspective, we argue that MLPs can also be made more accurate than their training methods non-empirically by generalizing across a third space, that of quantum-chemical methods. We propose routes to method-transferable MLPs that predict at unseen methods, with working implementations in both density-functional and wavefunction-based theories showing that they can be more accurate than any of their training methods. Aiming beyond the accuracy of the training methods could also make universal MLPs more accurate than single-method models, even when the method-transferable MLPs remain less accurate than those QM methods. Method-transferable MLPs will shape the construction of training datasets and the evaluation of models, pushing the accuracy limits of chemical simulations.
Related
- How accurate are foundational machine learning interatomic potentials for heterogeneous catalysis?
- Projected Hessian learning: fast curvature supervision for accurate machine-learning interatomic potentials
- Accurate and Rapid Vibrational Spectra of Molecular Crystals from Machine-Learned Interatomic Potentials
- FACTO: fast atom-centered tensor operator for accurate and efficient foundational machine-learned interatomic potentials
- Accurate, transferable, and verifiable machine-learned interatomic potentials for layered materials
BibTeX
@misc{dral2026machine,
title = {{Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On?}},
author = {Pavlo O. Dral and Yi-Fan Hou and Yuxinxin Chen and Hassan Nawaz and Mateusz Witkowski and Szymon Śmiga and Ireneusz Grabowski},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.18027},
eprint = {2610.18027},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.18027},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Can Machine Learning Potentials Be More Accurate Than the Methods They Are Trained On?
AU - Pavlo O. Dral
AU - Yi-Fan Hou
AU - Yuxinxin Chen
AU - Hassan Nawaz
AU - Mateusz Witkowski
AU - Szymon Śmiga
AU - Ireneusz Grabowski
PY - 2026
DA - 2026/10/13
PB - arXiv
DO - 10.48550/arxiv.2610.18027
UR - https://arxiv.org/abs/2610.18027
AB - Although machine learning potentials (MLPs) generalize across configurational and chemical compound spaces, their accuracy is commonly believed to be limited by their training methods, with improvements beyond this limit achieved so far only by refitting to experimental observables. In this Perspective, we argue that MLPs can also be made more accurate than their training methods non-empirically by generalizing across a third space, that of quantum-chemical methods. We propose routes to method-transferable MLPs that predict at unseen methods, with working implementations in both density-functional and wavefunction-based theories showing that they can be more accurate than any of their training methods. Aiming beyond the accuracy of the training methods could also make universal MLPs more accurate than single-method models, even when the method-transferable MLPs remain less accurate than those QM methods. Method-transferable MLPs will shape the construction of training datasets and the evaluation of models, pushing the accuracy limits of chemical simulations.
ER -