Organizations: School of Aeronautics and Astronautics, Zhejiang University, Hangzhou 310027, China · Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China · School of Physical Sciences, University of Chinese Academy of Science, Beijing 100049, China · College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
Liquid scintillator detectors are widely used in neutrino experiments due to their low energy threshold and high energy resolution. Despite the tiny abundance of 14C in LS, the photons induced by the β decay of the 14C isotope inevitably contaminate the signal, degrading the energy resolution. In this work, we propose three models to tag 14C photon hits in e+ events with 14C pile-up, thereby suppressing its impact on the energy resolution at the hit level: a gated spatiotemporal graph neural network and two Transformer-based models with scalar and vector charge encoding. For a simulation dataset in which each event contains one 14C and one e+ with kinetic energy below 5 MeV, the models achieve 14C recall rates of 25%-48% while maintaining e+ to 14C misidentification below 1%, leading to a large improvement in the resolution of total charge for events where e+ and 14C photon hits strongly overlap in space and time.
Figures & tables
Figure 1: Hit time distributions of pile-up events. The curves show the contributions from e+ annihilation (blue), 14C decay (red), and dark noise (gray), along with the total hit count (dashed black). (a) With large Δt , the 14C signal appears as a weak secondary peak. (b) With small Δt , the 14C hits are buried within the dominant e+ peak.
Figure 2: The spatiotemporal structure of a pile-up event with Δt=1.9ns . The truth information for the e+ event consists of a kinetic energy Ek(e+)=1.4MeV and a vertex position of (x,y,z)=(−1.9,5.7,11.1)m , while that for the 14C event consists of a deposited energy E14C=107.4keV and a vertex position of (x,y,z)=(−5.4,1.9,−14.8)m . The plots show the spatial distribution of photon hits in 100ns intervals. Blue dots represent hits from the positron annihilation, red dots from the 14C decay, and gray dots indicate dark noise.
Figure 3: Architecture of STT-Vector. Each hit is processed through two parallel encoding branches. The spatiotemporal branch (top) maps normalized spatial coordinates and hit times to a high-dimensional representation using sinusoidal positional encodings ψ(⋅) . The vector charge branch (bottom) embeds the 18-dimensional charge feature vector qi , which summarizes multi-scale local and global charge accumulation, using a dedicated multilayer perceptron. The two embeddings are concatenated, projected to the Transformer model dimension dmodel , and passed through L Transformer encoder layers to produce per-hit class logits for dark noise, e+ , and 14C .
Figure 4: A visualization of the hit identification results of Gated-STGNN, STT-Scalar, and STT-Vector, compared with the truth for a pile-up event with Δt=165.0ns . The truth information for the e+ event consists of a kinetic energy Ek(e+)=0MeV and a vertex position of (x,y,z)=(0.03,−0.02,−0.08)m , while that for the 14C event consists of a deposited energy E14C=82.5keV and a vertex position of (x,y,z)=(7.65,5.75,11.23)m . The plots display the (θ,ϕ,t) distribution of the hits, where t is the hit time and (θ,ϕ) are the polar and azimuthal angles of the PMT. Blue dots represent hits from the e+ event, red dots from the 14C decay, and gray dots indicate dark noise.
Figure 5: Normalized confusion matrices for events with z=0 and Ek(e+)=0MeV . The panels display the classification performance for Gated-STGNN, STT-Scalar, and STT-Vector.
Figure 6: Normalized confusion matrices for events with z=0 and Ek(e+)=5MeV . The panels display the classification performance for Gated-STGNN, STT-Scalar, and STT-Vector.
Figure 7: Classification metrics as functions of positron kinetic energy Ek(e+) at z=0m . The panels show the dependence on Ek(e+) of the fraction of e+ hits misidentified as 14C , the fraction of dark noise (DN) hits misidentified as 14C , and the 14C recall.
Figure 8: Classification metrics as functions of the vertex position z (evaluated at discrete points z∈{0,6,15,16.5}m ) for events with a fixed positron kinetic energy Ek(e+)=0MeV . The panels display the dependence on z of the fraction of e+ hits misidentified as 14C , the fraction of dark noise (DN) hits misidentified as 14C , and the 14C recall.
Figure 9: Differential 14C recall R14C(Δt,E14C) of STT-Vector, evaluated at z=0 for Ek(e+)=0MeV and Ek(e+)=5MeV , respectively.
Figure 10: Distributions of the total charge of selected hits ( Qsum ) at the detector center for events with Ek(e+)=0MeV (left) and Ek(e+)=5MeV (right). The gray dotted line represents Qsum of all hits, before applying hits removal. The green dashed line shows the ideal distribution with 14C pile-up removed by truth label. The solid purple line represents Qsum after hit removal based on the STT-Vector model prediction. The fitted mean value ( μ ), width ( σ ) and corresponding resolution ( R=σ/μ ) are also shown.
Figure 11: Ratios of the resolutions of the charge sum of selected hits ( Qsum ) under different hit-removal criteria relative to the “Total” case, in which no hit is removed. The left plot shows the ratio’s dependence on Ek at the detector center. The right plot shows the ratio’s dependence on z when Ek(e+)=0MeV .
Scintillation detectors with excellent timing resolution enable more precise localization of radiation sources in positron emission tomography, leading to substantial improvements in diagnostic capability for diseases such as cancer and dementia. At the extreme timing precision required for such applications at the picosecond scale, detector performance is governed by the microscopic dynamics of scintillation photons generated within the detector and their subsequent detection processes. However, detector signals have conventionally been treated only as collective responses of many photons due to structural constraints inherent to photodetectors. In this study, we overcome this fundamental limitation using deep learning, enabling direct access to the timing information of individual photons. The proposed method estimates photon-by-photon arrival times directly from detector waveforms without requiring any modification to the detector structure; the method operates on an event-by-event basis without ground-truth labels by integrating an unsupervised learning framework with a physically informed detector-response model. Through comprehensive validation combining Monte Carlo simulation and experimental measurements across various detector configurations, we experimentally demonstrate improved timing resolution, visualized depth-of-interaction-dependent photon transport, and classified Cherenkov and scintillation photons based on the estimated photon-level timing information using a unified deep learning-based framework. These results provide experimental access to photon dynamics, bridging the gap between theoretical modeling and experimental observation, and they open a new data-driven pathway for discovery in detector physics and optimization.
Yuya Onishi, Ryosuke Ota, Fumio Hashimoto +4
Central Research Laboratory, Hamamatsu Photonics K.K. · Graduate School of Science and Engineering, Chiba University · Department of Advanced Nuclear Medicine Sciences, National Institutes for Quantum Science and Technology +1
Modern machine learning techniques have become increasingly important in particle physics because of their powerful pattern-recognition capabilities, including in real-time data acquisition where stringent runtime constraints apply. This paper details the performance of deep-learning-based trigger algorithms for a large water Cherenkov detector such as Hyper-Kamiokande, aimed at low-energy neutrino events (below 7 MeV). The performance of custom neural-network supervised classifiers is shown alongside two anomaly-detection approaches trained solely on detector noise: a pure autoencoder and a model based on Manifold Projection-Diffusion Recovery. The supervised model shows signal identification efficiencies of 76.7% for single electrons of 3 MeV kinetic energy, significantly exceeding signal efficiencies obtained from a traditional hit-count-based trigger of 26.4%, while the Manifold Projection-Diffusion Recovery approach reaches 35.4% at the same operating point. Runtime evaluations on GPU yield per-window inference latencies well below the millisecond scale.
Katharina Lachner, Saúl Alonso-Monsalve, Benjamin Richards +1
Charged-particle tracking -- reconstructing trajectories from sparse detector measurements -- is a fundamental high-energy-physics inference problem and a canonical example of learning under extreme combinatorial ambiguity. At the High-Luminosity Large Hadron Collider (HL-LHC), tracking must remain accurate and efficient despite unprecedented collision densities. Graph neural networks perform strongly, but incur substantial costs from graph construction and processing, while transformer-based approaches rely on auxiliary stages that prevent end-to-end optimization. To address this, we present HEPTv2, an end-to-end point-transformer architecture that reconstructs tracks from detector hits in one trainable pipeline. HEPTv2 combines a locality-aware point encoder with a track decoder that predicts complete trajectories without graph-building, clustering, or filtering. The encoder uses locality-sensitive hashing in detector coordinate space to preserve tracking-relevant geometry while enabling efficient local attention. The decoder resolves ambiguities through sectorized decoding and direct hit-to-track prediction under joint encoder-decoder supervision, allowing the full pipeline to be optimized end-to-end. On TrackML, HEPTv2 achieves 98.6% double-majority tracking efficiency at a 0.8% fake rate, while requiring only ∼15ms inference time and 0.4GB peak memory per event on a NVIDIA A100 GPU. Latency and memory scale approximately linearly for events with up to 5×105 hits. HEPTv2 establishes a new state of the art in the accuracy-latency trade-off, improving efficiency by 4.5% over the strongest prior transformer and by 1.1--2.2% over optimized graph-based pipelines, while reducing latency by factors of 7 and 38--52, respectively. These results show end-to-end transformers can deliver the accuracy and efficiency required for real-time particle reconstruction at the HL-LHC.
Siqi Miao, Shitij Govil, Jack P. Rodgers +5
School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA. · Department of Physics and Astronomy, Purdue University, West Lafayette, IN, USA. · Department of Physics, University of California San Diego, La Jolla, CA, USA. +1