Domain Adaptive Graph Neural Networks for Neutrino Experiments
Authors: Hemalata Nayak, Giuseppe Cerati, Aleksandra Ćiprijanović, Tammy Walton
Organizations: School of Computing, Data Sciences & Physics The College of William & Mary Williamsburg, VA 23185 · Scientific Computing Division Fermi National Accelerator Laboratory Batavia, IL 60510 · Consortium for Advanced Science and Engineering (CASE) University of Chicago, Chicago, IL 60637 · Department of Astronomy & Astrophysics University of Chicago, Chicago, IL 60637 · NSF-Simons AI Institute for the Sky (SkAI) Chicago, IL 60611
Weakly interacting particles such as neutrinos require highly efficient and precise detection methods to enable searches for new physics. Deep learning, and particularly Graph Neural Networks (GNNs), has shown promise for reconstruction tasks in liquid argon time projection chamber (LArTPC) neutrino experiments. However, differences between simulations and experimental data introduce domain shifts that can degrade model performance and potentially mimic or obscure new-physics signals. Here, we demonstrate that robust learning across datasets can still be achieved by using domain adaptation during training. We introduce, to our knowledge for the first time, domain-adaptation-enabled GNNs for robust LArTPC reconstruction tasks. We use MicroBooNE open dataset as our labeled source domain, and introduce four perturbed datasets as our unlabeled target domains-- including a fixed scale variation, moderate and severe gaussian noise, and a signal variation as a function of the position in the detector. The inclusion of adversarial-based domain adaptation during training improves performance for all four target domains, without a substantial drop in performance in the source domain. Precision in the target domain was improved up to 0.19 and recall up to 0.37, depending on the variation and the network task. These results demonstrate a potential of domain adaptation to bridge the gap between simulations and experiments, and build more trustworthy models for current and future LArTPC neutrino experiments.
Figures & tables
Semantic–Source
Semantic–Target
Filter–Source
Filter–Target
Experiment
No-DA
DA
No-DA
DA
No-DA(P/R)
DA(P/R)
No-DA(P/R)
DA(P/R)
DA 1
0.89
0.87
0.68
0.83
0.96 / 0.97
0.95 / 0.97
0.93 / 0.56
0.93 / 0.93
DA 2
0.90
0.90
0.86
0.89
0.96 / 0.97
0.96 / 0.97
0.96 / 0.94
0.96 / 0.96
DA 3
0.91
0.88
0.67
0.86
0.96 / 0.97
0.95 / 0.97
0.96 / 0.80
0.95 / 0.95
DA 4
0.89
0.90
0.86
0.89
0.96 / 0.97
0.96 / 0.97
0.97 / 0.87
0.97 / 0.94
Table 1: Precision (P) and recall (R) of the semantic and filter decoders evaluated on the source and target test sets. Due to training with the recall loss [ 24 ] , these are by construction the same value for the semantic decoder, while for the filter decoder, we report them as (P/R). For each experiment ( DA1 – DA4 ), decoder (semantic or filter), and domain (source or target), the better-performing result between the No-DA and DA is shown in bold. Results have a statistical precision of O(1%) or less, estimated as the standard deviation of five repetitions of the training with different random seeds.
Figure 1: Confusion matrices showing precision (top) and recall (bottom) of the semantic decoder, when models are trained without (left - source domain; middle - target domain), and with DA (right - target domain). Both matrices are obtained from the same inputs: for precision the matrix is normalized by column (assigned labels), while for recall by row (true label). Results are shown for test sets of time-dependent RMS shift ( DA4 ). Inclusion of DA improves both precision and recall.
Figure 2: Same as Figure 1 , but for filter decoder.
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Figure 3: Latent space embeddings (plotted using UMAP [ 17 ] plot with 10 neighbors for each point) for models trained without (left) and with DA (right), in DA4 experiments. Source domain (circles), and target domain (triangles) data with two filter decoder classes (noise in red, and signal in blue) are shown. Inclusion of DA improves alignment and helps slightly improve separation between the two classes.
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.
Ivan Kharuk
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 +1
We present a neural-network-based data processing pipeline for Baikal-GVD, designed to improve event reconstruction quality and accelerate neutrino candidates selection. The pipeline comprises three stages: fast suppression of extensive air shower events, suppression of noise optical modules activations, and extraction of high confidence neutrino candidates. All three networks employ a transformer architecture that exploits inter-hit correlations through the attention mechanism. Applied sequentially, the pipeline achieves orders-of-magnitude speedup over the standard reconstruction chain. Moreover, noise suppression neural network surpasses the accuracy of algorithmic noise suppression algorithms and provides estimate for time residuals of the signal hits, which is crucial for identification of track-like hits. We address the domain shift between Monte Carlo simulations and experimental data by incorporating a domain adaptation technique, demonstrating improved agreement between the two domains. The resulting framework enables near-real-time event classification, with direct applications to multi-messenger alert systems and diffuse neutrino flux measurements.
A. Matseiko, G. Plotnikov, I. Kharuk
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 +1
In the forthcoming years the LHC experiments are going to be upgraded to benefit from the substantial increase of the LHC instantaneous luminosity, which will lead to larger, denser events, and, consequently, greater complexity in reconstructing charged particle tracks, motivating frontier research in new technologies. Quantum machine learning models are being investigated as potential new approaches to high energy physics (HEP) tasks. We characterize and upgrade a quantum graph neural network (QGNN) architecture for charged particle track reconstruction on a simulated high luminosity dataset. The model operates on a set of event graphs, each built from the hits generated in tracking detector layers by particles produced in proton collisions, performing a classification of the possible hit connections between adjacent layers. In this approach the QGNN is designed as a hybrid architecture, interleaving classical feedforward networks with parametrized quantum circuits. We characterize the interplay between the classical and quantum components. We report on the principal upgrades to the original design, and present new evidence of improved training behavior, specifically in terms of convergence toward the final trained configuration.
Matteo Argenton, Laura Cappelli, Concezio Bozzi
Department of Physics and Earth Science, University of Ferrara, Via Saragat 1, Ferrara, 44122, Italy · INFN Ferrara, Via Saragat 1, Ferrara, 44122, Italy