Safe Domain Adaptation for Physics: Overcoming Nuisances, Label Shifts, and Simulation Priors

Ivan Kharuk

cs.LGAug 18, 2026

Safe Domain Adaptation for Physics: Overcoming Nuisances, Label Shifts, and Simulation Priors

Authors: Ivan Kharuk

Organizations: Institute for Nuclear Research of the Russian Academy of Sciences · Moscow Institute of Physics and Technology · Institute for Nuclear Research of the Russian Academy of Sciences, 60th October Anniversary Prospect 7a, Moscow, 117312, Russia · Moscow Institute of Physics and Technology, Institutsky lane, 9, Dolgoprudny, Moscow region, 141700, Russia

Abstract

Domain adaptation is widely used to make neural networks trained on simulations applicable to experimental data. Its premise is that the two domains differ only in nuisances, and that the quantity of interest is distributed identically in both. In physics neither assumption holds: simulations can be wrong about the physics, and the distribution of the target quantity - an energy spectrum, a redshift distribution - is often the measurement itself. We study the consequences of such mismatches on a toy air-shower benchmark in which a detector-response nuisance, a physical simulation shift, and an energy-spectrum shift can be switched on separately or together. Standard adversarial adaptation handles the conditional shifts, but once the two spectra differ it aligns them, replacing an uncontrolled bias by one anchored on the simulation prior. We present adaptive domain adaptation, which reweights the simulated events so as to focus domain adaptation on the genuine physical mismatch alone. Since the predicted spectrum depends on model training configuration, we provide a label-free model selection rule for selecting the near-the-best operation point.

Figures & tables

Panel of Figure 1 — Figure 1: An example of an event. (a): station locations and registered signal, (b): example of two waveforms registered by stations, with time axis in arbitrary units (a.u.).

Figure 2: Data flow during training and the network architecture. The labelled training set

\mathcal{D}_{\mathrm{tr}}

forms the regression batch (blue), which is processed by the transformer encoder and read by the regression head. The DA-source

\mathcal{D}_{\mathrm{da}}

and the target

\mathcal{D}_{\mathrm{tgt}}

are combined in equal proportion into the domain-adaptation batch (orange), which is processed by the same encoder and read by the domain-adaptation layer. The two paths meet in the combined loss.

Panel of Figure 3 — Figure 3: (a) The true target spectrum, $\mathrm{Beta}(2,5)$ , and the $\mathrm{Beta}(4,6)$ spectrum used for the DA-source where a label shift is present. (b) A source (Gaussian) and target (Laplacian) waveform carrying the same integrated charge but a different pulse shape. (c) The true target spectrum and the target spectrum reconstructed without domain adaptation (baseline) and with DANN.

Panel of Figure 4 — Figure 4: Pure lateral-distribution mismatch. (a) The predicted target spectrum with no adaptation and with DANN at the adopted recipe, against the true $\mathrm{Beta}(2,5)$ target. (b) Per-seed predicted target spectra at the optimal $\lambda$ and at a poorly-chosen (too-large) $\lambda$ .

Panel of Figure 5 — Figure 5: Reconstructed target spectra against the true $\mathrm{Beta}(2,5)$ target. (a) Waveform and LDF mismatch with matched spectra, DA-source and target both $\mathrm{Beta}(2,5)$ : single event critic and disentangled two-critic scheme. (b, c) LDF and spectrum mismatch, true target $\mathrm{Beta}(2,5)$ and DA-source $\mathrm{Beta}(3,5)$ , at LDF slope $+0.5E$ (b) and $-0.2E$ (c), comparing ADA and DANN predictions.

Panel of Figure 6 — Figure 6: Target per-event error $\mathrm{MAE}_{\mathrm{tgt}}$ against the DA-source $-$ target spectrum-mean difference, for event ADA and DANN with the no-adaptation baseline, at LDF slope (a) $+0.5E$ , (b) $+0.2E$ and (c) $-0.2E$ . Each operating point is chosen by the label-free rule of section 5 ; bands are $\pm 1$ std over the seeds.

Panel of Figure 7 — Figure 7: Waveform-plus-spectrum mismatch for event ADA, DANN, staged DA, and waveform ADA with the no-adaptation baseline, against the difference between the DA-source and target spectrum means: (a) the target recovery $\mathrm{WD}_{\mathrm{tgt}}$ and (b) the calibration slope $\kappa$ .

Panel of Figure 8 — Figure 8: Iterative domain adaptation applied to staged DA. (a) Target recovery $\mathrm{WD}_{\mathrm{tgt}}$ against the iteration round, for three DA-source spectra at mean differences $+0.11$ , $+0.21$ and $+0.31$ . (b) Predicted target spectra across rounds for the largest mismatch, $\Delta=+0.31$ .

Panel of Figure 9 — Figure 9: Target recovery $\mathrm{WD}_{\mathrm{tgt}}$ against the DA-source—target spectrum-mean difference for event ADA and DANN for the triple mismatch at LDF slope (a) $+0.2E$ and (b) $-0.2E$ .

Panel of Figure 10 — Figure 10: Iterative ADA under the triple mismatch. (a) Target recovery $\mathrm{WD}_{\mathrm{tgt}}$ against the iteration round, for a badly-mismatched prior ( $\alpha=-0.2$ , DA-source $\mathrm{Beta}(3,2)$ ) and a mild one ( $\alpha=+0.2$ , flat DA-source); round 1 is single-pass ADA and the dotted lines are the corresponding no-adaptation baselines. (b) Predicted target spectra across rounds for the $-0.2E$ case, converging from the initial prior towards the true $\mathrm{Beta}(2,5)$ .

candidate label-free signal	kind	$\rho$
marginal separability $\mathrm{LR}_{\mathrm{acc}}$	marginal latent	$+0.3$
conditional discrepancy (CMMD)	conditional latent	$+0.77$
augmentation consistency	prediction stability	$-0.4$ to $-0.7$
cross-seed spread	prediction stability	$\mathbf{+1.00}$

Table 1: Label-free selection signals ranked (Spearman

\rho

) against the true target recovery

\mathrm{WD}_{\mathrm{tgt}}

on the waveform-and-LDF case of section 4.1 , oriented so that a positive value means the signal ranks recovery correctly. The no-DA baseline is excluded from the ranking.

knowledge of the target spectrum	domain shift	recommended scheme
—	none detected	no adaptation; the network already interpolates
trustworthy	any	DANN with an event-wise critic
approximate	localized	staged DA or ADA with localized critic
approximate	not localized	ADA with an event-wise critic

Table 2: Practical guidance on the choice of the adaptation scheme.

Domain Adaptation

Domain adaptation transfers machine-learning models across changing data distributions, often using unlabeled or limited target-domain examples.

See all 252 papers

Unsupervised Domain Adaptation for Multitask Image Analysis in Realistic Context with Extreme Label Shift; Application to the CTAO first Large Sized Telescope
Aug 10, 2026Michaël Dell'aiera, Thomas Vuillaume, Alexandre BenoitClass-Imbalanced Learning Label Shift
CNRS LAPP, Annecy, 74940, France · Université Savoie Mont Blanc LISTIC, Annecy, 74940, France
PhysGuard: Fisher-Guided Gradient Projection for Sim-to-Real Neural PDE Surrogates
Jun 15, 2026Changjian Zhou, Junfeng Fang, Negin Yousefpour +3Synthetic-to-Real Domain Adaptation PDE Surrogate Modeling
Faculty of Engineering and IT, University of Melbourne · School of Computing, National University of Singapore · Artificial Intelligence Research Institute, IFLYTEK Co., Ltd.
Bridging the Simulation-to-Experiment Gap with Generative Models using Adversarial Distribution Alignment
Apr 1, 2026Kai Nelson, Tobias Kreiman, Sergey Levine +1Generative Modeling Synthetic-to-Real Domain Adaptation
University of California, Berkeley · Lawrence Berkeley National Laboratory (LBNL)

Aug 10, 2026J/K move · Enter open · S save

Unsupervised Domain Adaptation for Multitask Image Analysis in Realistic Context with Extreme Label Shift; Application to the CTAO first Large Sized Telescope

Michaël Dell'aiera, Thomas Vuillaume, Alexandre Benoit

CNRS LAPP, Annecy, 74940, France · Université Savoie Mont Blanc LISTIC, Annecy, 74940, France

Class-Imbalanced Learning Label Shift Unsupervised Domain Adaptation Multi-Task Learning Importance Weighting

Open paper AskPDF arXiv

Unsupervised domain adaptation is a widespread set of methods that leverages the knowledge of a labeled source domain to train a model to perform well on a related unlabeled target domain. They generally introduce an auxiliary adaptation-related task that can be integrated into the multitask paradigm, which aims to merge multiple single-task models into a unified architecture. In this paper, we propose to associate domain adaptation and multitask balancing in the realistic context of an extreme class imbalance. Therefore, we propose a combined framework to cover and validate these approaches, and evaluate its performance in the physics-based context of the Cherenkov Telescope Array Observatory (CTAO). Along with a comparative study of some relevant adaptation techniques, we highlight the impact of extreme label shift and extend the investigations on importance weighting to rectify it. The complete code and results are published and available as open-source resources on Zenodo.

Abstract

Domain Adaptation

Explore similar work

Unsupervised Domain Adaptation for Multitask Image Analysis in Realistic Context with Extreme Label Shift; Application to the CTAO first Large Sized Telescope

PhysGuard: Fisher-Guided Gradient Projection for Sim-to-Real Neural PDE Surrogates

Bridging the Simulation-to-Experiment Gap with Generative Models using Adversarial Distribution Alignment