Faithful Uncertainties in ML-Based Inverse Problems: GPs, NNs and NTKs for the Triplet PDF
Abstract
Machine-learning methods are increasingly used to solve inverse problems in physics, where a faithful estimate of the uncertainty on the reconstructed function is as important as its central value. We address this question in the simplest instance of a parton distribution function (PDF) determination: the reconstruction of the non-singlet triplet distribution T3(x,Q02) from deep-inelastic scattering data. Within a closure-test framework, in which the underlying PDF is known and the pseudo-data carry correlated experimental uncertainties, we compare Gaussian processes with different kernels, neural-network ensembles trained with different loss functions, and empirical neural tangent kernel approximations. We find that methods yielding comparable central reconstructions can produce markedly different uncertainties, and we identify the ingredients that control them: a likelihood that includes the full experimental covariance, the functional prior encoded in the kernel or in the endpoint preprocessing, and, for tangent-kernel approximations, the stage of training at which the kernel is evaluated. A normalised-bias diagnostic shows that calibration depends on the correlations of the predictive covariance, not only on its pointwise width. Our results illustrate why closure tests and covariance-aware diagnostics are essential to validate uncertainties in ML-based approaches to inverse problems.
Related
- Physical Fidelity of Wave-Based Inverse Problems under Broken Time-Reversal Symmetry
- Bayesian bilevel operator learning with low-rank adaptation for efficient uncertainty quantification of PDE inverse problems.
- Benchmarking Optimizers to Solve Inverse Problems with Differentiable Physics Simulators
- iPINN for Broadband CARS Phase Retrieval: A Framework for Function Approximation and Inverse Modeling Problems in Nonlinear Spectroscopy
- Scalable Bayesian Optimization of Composite Functions for Image-Based Inverse Problems in Materials Characterization
BibTeX
@misc{elmer2026faithful,
title = {{Faithful Uncertainties in ML-Based Inverse Problems: GPs, NNs and NTKs for the Triplet PDF}},
author = {Nina Elmer and Sven Krippendorf and Maria Ubiali},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.30239},
eprint = {2610.30239},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.30239},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Faithful Uncertainties in ML-Based Inverse Problems: GPs, NNs and NTKs for the Triplet PDF
AU - Nina Elmer
AU - Sven Krippendorf
AU - Maria Ubiali
PY - 2026
DA - 2026/10/21
PB - arXiv
DO - 10.48550/arxiv.2610.30239
UR - https://arxiv.org/abs/2610.30239
AB - Machine-learning methods are increasingly used to solve inverse problems in physics, where a faithful estimate of the uncertainty on the reconstructed function is as important as its central value. We address this question in the simplest instance of a parton distribution function (PDF) determination: the reconstruction of the non-singlet triplet distribution T_3(x,Q_0^2) from deep-inelastic scattering data. Within a closure-test framework, in which the underlying PDF is known and the pseudo-data carry correlated experimental uncertainties, we compare Gaussian processes with different kernels, neural-network ensembles trained with different loss functions, and empirical neural tangent kernel approximations. We find that methods yielding comparable central reconstructions can produce markedly different uncertainties, and we identify the ingredients that control them: a likelihood that includes the full experimental covariance, the functional prior encoded in the kernel or in the endpoint preprocessing, and, for tangent-kernel approximations, the stage of training at which the kernel is evaluated. A normalised-bias diagnostic shows that calibration depends on the correlations of the predictive covariance, not only on its pointwise width. Our results illustrate why closure tests and covariance-aware diagnostics are essential to validate uncertainties in ML-based approaches to inverse problems.
ER -