Uncertainty quantification for neural network potential foundation models
Abstract For neural network potentials (NNPs) to gain widespread use, researchers must be able to trust model outputs. However, the blackbox nature of neural networks and their inherent stochasticity are often deterrents, especially for foundation models trained over broad swaths of chemical space....
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-04-01
|
| Series: | npj Computational Materials |
| Online Access: | https://doi.org/10.1038/s41524-025-01572-y |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849709658788331520 |
|---|---|
| author | Jenna A. Bilbrey Jesun S. Firoz Mal-Soon Lee Sutanay Choudhury |
| author_facet | Jenna A. Bilbrey Jesun S. Firoz Mal-Soon Lee Sutanay Choudhury |
| author_sort | Jenna A. Bilbrey |
| collection | DOAJ |
| description | Abstract For neural network potentials (NNPs) to gain widespread use, researchers must be able to trust model outputs. However, the blackbox nature of neural networks and their inherent stochasticity are often deterrents, especially for foundation models trained over broad swaths of chemical space. Uncertainty information provided at the time of prediction can help reduce aversion to NNPs. In this work, we detail two uncertainty quantification (UQ) methods. Readout ensembling, by finetuning the readout layers of an ensemble of foundation models, provides information about model uncertainty, while quantile regression, by replacing point predictions with distributional predictions, provides information about uncertainty within the underlying training data. We demonstrate our approach with the MACE-MP-0 model, applying UQ to the foundation model and a series of finetuned models. The uncertainties produced by the readout ensemble and quantile methods are demonstrated to be distinct measures by which the quality of the NNP output can be judged. |
| format | Article |
| id | doaj-art-2e531f44a0b74b3fae4f02196a1b6a74 |
| institution | DOAJ |
| issn | 2057-3960 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Computational Materials |
| spelling | doaj-art-2e531f44a0b74b3fae4f02196a1b6a742025-08-20T03:15:12ZengNature Portfolionpj Computational Materials2057-39602025-04-011111810.1038/s41524-025-01572-yUncertainty quantification for neural network potential foundation modelsJenna A. Bilbrey0Jesun S. Firoz1Mal-Soon Lee2Sutanay Choudhury3AI & Data Analytics, Pacific Northwest National LaboratoryAdvanced Computing, Mathematics, & Data, Pacific Northwest National LaboratoryChemical Physics & Analysis, Pacific Northwest National LaboratoryAdvanced Computing, Mathematics, & Data, Pacific Northwest National LaboratoryAbstract For neural network potentials (NNPs) to gain widespread use, researchers must be able to trust model outputs. However, the blackbox nature of neural networks and their inherent stochasticity are often deterrents, especially for foundation models trained over broad swaths of chemical space. Uncertainty information provided at the time of prediction can help reduce aversion to NNPs. In this work, we detail two uncertainty quantification (UQ) methods. Readout ensembling, by finetuning the readout layers of an ensemble of foundation models, provides information about model uncertainty, while quantile regression, by replacing point predictions with distributional predictions, provides information about uncertainty within the underlying training data. We demonstrate our approach with the MACE-MP-0 model, applying UQ to the foundation model and a series of finetuned models. The uncertainties produced by the readout ensemble and quantile methods are demonstrated to be distinct measures by which the quality of the NNP output can be judged.https://doi.org/10.1038/s41524-025-01572-y |
| spellingShingle | Jenna A. Bilbrey Jesun S. Firoz Mal-Soon Lee Sutanay Choudhury Uncertainty quantification for neural network potential foundation models npj Computational Materials |
| title | Uncertainty quantification for neural network potential foundation models |
| title_full | Uncertainty quantification for neural network potential foundation models |
| title_fullStr | Uncertainty quantification for neural network potential foundation models |
| title_full_unstemmed | Uncertainty quantification for neural network potential foundation models |
| title_short | Uncertainty quantification for neural network potential foundation models |
| title_sort | uncertainty quantification for neural network potential foundation models |
| url | https://doi.org/10.1038/s41524-025-01572-y |
| work_keys_str_mv | AT jennaabilbrey uncertaintyquantificationforneuralnetworkpotentialfoundationmodels AT jesunsfiroz uncertaintyquantificationforneuralnetworkpotentialfoundationmodels AT malsoonlee uncertaintyquantificationforneuralnetworkpotentialfoundationmodels AT sutanaychoudhury uncertaintyquantificationforneuralnetworkpotentialfoundationmodels |