AdaptNTK: Adaptive Uncertainty Quantification and Active Learning for Neural Network Potentials

Abstract

Machine learning interatomic potentials bridge the gap between quantum chemical precision and classical computational speed, enabling molecular dynamics simulations with first-principles accuracy. Their reliability is often improved through active learning, which iteratively expands the training set by identifying uncertain, out-of-distribution configurations. Existing uncertainty-quantification methods often involve a trade-off between computational cost and reliability, and generally cannot account for redundancy as an acquisition batch is assembled. Here, we introduce AdaptNTK, a single-model framework that measures uncertainty as a regularized Mahalanobis distance in empirical neural tangent kernel (NTK) feature space. With the NTK features fixed during acquisition, the uncertainty depends on the acquired configurations but not their reference labels. This allows the uncertainty to be updated recursively after each selection without retraining, reducing redundancy within an acquisition batch. On held-out rMD17 data, AdaptNTK achieves the highest mean correlations with force errors (Spearman 0.68, Pearson 0.71) and matches a three-member ensemble in error retention. In active learning experiments, AdaptNTK achieves the lowest force errors across rMD17 and Transition-1X, with particularly strong performance on transition-state configurations in Transition-1X. AdaptNTK provides a 2.6-fold speedup per Transition-1X cycle relative to the ensemble, providing efficient single-model uncertainty estimation with sequential updates for data-efficient active learning.

Bayesian neural networks versus deep ensembles for uncertainty quantification in machine learning interatomic potentials
Riccardo Farris, Emanuele Telari, Nongnuch Artrith +2
Machine Learning Science and Technology24 Aug 2026Peer-reviewedCC BYFull text6 citedMaterials & polymersOpenAlex
Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
Stefan P. Schmid, Tobias Setzer, Maximilian Scheurer
arXiv30 Oct 2026PreprintCC BY 4.0Full textPhysical chemistryarXiv
Comparative study of ensemble-based uncertainty quantification methods for neural network interatomic potentials
Yonatan Kurniawan, Mingjian Wen, Ellad B. Tadmor +1
Machine Learning Science and Technology27 Aug 2026Peer-reviewedCC BY6 citedMaterials & polymersOpenAlex
NephroTox AI: an explainable machine learning framework with uncertainty quantification for drug-induced nephrotoxicity prediction
Jixing Liu, Jiabing Guo, Henry H. Y. Tong +1
Journal of Cheminformatics5 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Theoretical & computational chemistryCrossref
Error Models for Uncertainty Quantification in Molecular Machine Learning
Raquel Parrondo-Pizarro, Jessica Lanini, Raquel Rodríguez-Pérez
ChemRxiv31 Aug 2026PreprintCC BY-NC-NDFull textGeneral chemistryChemRxiv

Cite

BibTeX RIS

BibTeX

@misc{ananth2026adaptntk,
  title = {{AdaptNTK: Adaptive Uncertainty Quantification and Active Learning for Neural Network Potentials}},
  author = {Prajwal Ananth and Shuwen Yue},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.00488},
  eprint = {2609.00488},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.00488},
}

RIS

TY  - UNPB
TI  - AdaptNTK: Adaptive Uncertainty Quantification and Active Learning for Neural Network Potentials
AU  - Prajwal Ananth
AU  - Shuwen Yue
PY  - 2026
DA  - 2026/09/01
PB  - arXiv
DO  - 10.48550/arxiv.2609.00488
UR  - https://arxiv.org/abs/2609.00488
AB  - Machine learning interatomic potentials bridge the gap between quantum chemical precision and classical computational speed, enabling molecular dynamics simulations with first-principles accuracy. Their reliability is often improved through active learning, which iteratively expands the training set by identifying uncertain, out-of-distribution configurations. Existing uncertainty-quantification methods often involve a trade-off between computational cost and reliability, and generally cannot account for redundancy as an acquisition batch is assembled. Here, we introduce AdaptNTK, a single-model framework that measures uncertainty as a regularized Mahalanobis distance in empirical neural tangent kernel (NTK) feature space. With the NTK features fixed during acquisition, the uncertainty depends on the acquired configurations but not their reference labels. This allows the uncertainty to be updated recursively after each selection without retraining, reducing redundancy within an acquisition batch. On held-out rMD17 data, AdaptNTK achieves the highest mean correlations with force errors (Spearman 0.68, Pearson 0.71) and matches a three-member ensemble in error retention. In active learning experiments, AdaptNTK achieves the lowest force errors across rMD17 and Transition-1X, with particularly strong performance on transition-state configurations in Transition-1X. AdaptNTK provides a 2.6-fold speedup per Transition-1X cycle relative to the ensemble, providing efficient single-model uncertainty estimation with sequential updates for data-efficient active learning.
ER  -

Abstract

Related

Cite