Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

The paper claims that a physics-guided hypergraph transformer with a learnable pileup-suppression gate recovers jet energy and mass under extreme HL-LHC pileup better than all eight tested baselines, and reconstructs the top-quark mass reso

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 21:19 UTC pith:GQTLYWBW

load-bearing objection A useful public benchmark and a well-described architecture, but the 'consistently outperforms' claim needs validation-set tuning and error bars before it is credible. the 4 major comments →

arxiv 2602.20475 v2 pith:GQTLYWBW submitted 2026-02-24 hep-ex cs.LG

PhyGHT: Physics-Guided HyperGraph Transformer for Signal Purification at the HL-LHC

Mohammed Rakib , Luke Vaughan , Shivang Patel , Flera Rizatdinova , Alexander Khanov , Atriya Sen This is my paper
classification hep-ex cs.LG
keywords pileup mitigationhypergraph transformergraph attentionphysics-guided machine learningtop quark mass reconstructionHL-LHCjet energy correctionopen simulated dataset
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper seeks to establish that a single neural architecture, PhyGHT, can remove contamination from up to two hundred simultaneous proton-proton collisions and recover the energy and mass of signal jets. It does this by predicting, for each jet, two bounded correction factors—the fraction of the raw jet energy and mass that originated from the hard-scatter collision—using a hybrid of local distance-aware graph attention, global transformer context, an explicit learnable soft-mask called the Pileup Suppression Gate, and hypergraph attention for aggregation. On a newly released simulated dataset of top-quark pair production, PhyGHT beats all eight baselines, including physics-based production algorithms and prior ML methods, with the largest gains in mass reconstruction; it also reconstructs the top-quark mass resonance in close agreement with truth. If the claims hold, pileup mitigation at the HL-LHC could become a two-number correction per jet, computed faster than existing approaches.

Core claim

The central discovery, on the paper's own benchmark, is that PhyGHT predicts per-jet energy and mass correction factors with test-set R^2 of 0.943 and 0.869 at a mean of 60 pileup collisions per event, and 0.932 and 0.836 at 200—consistently above all baselines. The decisive component, per the ablations, is the Pileup Suppression Gate: a differentiable soft-mask that estimates the probability each track comes from the hard-scatter vertex and multiplies that track's features by the probability before aggregation. Removing this gate costs several points of R^2 depending on the target, while removing the global transformer context costs more than any other single component. After applying the p

What carries the argument

The Pileup Suppression Gate (PSG) is the central object: a learnable, differentiable soft-mask that maps a track's fused local-and-global embedding to a scalar signal probability via a small MLP, then multiplies the embedding by that probability. It is trained with auxiliary track-level labels (hard-scatter vs pileup) so the mask tracks physical vertex association, and it is applied before the hypergraph aggregation stage. Around it, PhyGHT stacks a Distance-Aware Graph Attention layer that biases neighbor weights by a learnable spatial-decay parameter (a learned 'soft cone' in eta-phi space), a Transformer encoder for event-wide pileup density, and a bipartite/hypergraph attention that per

Load-bearing premise

The load-bearing premise is that the simulated, particle-level dataset—built by overlaying Poisson-distributed min-bias collisions, smearing vertices with Gaussians, and discarding neutral particles and low-pT tracks—captures the pileup contamination a real detector will see; if detector-level effects such as fake tracks, calorimeter response, and reconstruction inefficiencies break this proxy, the reported gains may not transfer to actual collision data.

What would settle it

A concrete check: evaluate PhyGHT and PUMINet on a full detector simulation (including calorimeter response, fake tracks, and reconstruction) of the same ttbar events; if PhyGHT's R^2 advantage over PUMINet shrinks below the reported margin or reverses, the central claim as stated for HL-LHC conditions would fail. A second check: rerun PUPPI with generator-level filtering on the hard-scatter events, as the original method did; if its mass R^2 at <mu>=200 rises substantially above the reported 0.114, the claim that PhyGHT outperforms the physics baseline is weakened.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the central claim is correct, pileup mitigation becomes a per-jet two-number correction: multiply the raw jet energy and mass by the predicted fractions, and downstream physics analyses proceed.
  • Because the gains hold as pileup rises from 60 to 200 (mass R^2 falls only 0.033), the method is positioned for the HL-LHC's extreme environment rather than just current conditions.
  • The top-quark mass reconstruction, a standard physics benchmark, shows the corrections are physically usable, not just good on a regression metric.
  • PhyGHT does this at lower inference latency than the strongest baselines (about 1.9x faster at <mu>=200), which matters for offline reprocessing of the enormous HL-LHC dataset.
  • The released simulated dataset and code allow independent reproduction, lowering the barrier for other groups to develop and compare pileup-mitigation methods.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper does not compare the PSG's learned track weights to PUPPI's physics-derived weights on the same events; a direct calibration study could show whether the learnable gate recovers the same signal/pileup discrimination or something qualitatively different.
  • Because the model predicts only two scalar fractions per jet, it discards information about how the hard-scatter energy is distributed within the jet; extending the same architecture to per-subjet or per-momentum-component corrections could further improve mass resolution and is a natural, untested follow-up.
  • The strong degradation of PUPPI at <mu>=200 (mass R^2 0.114) is flagged by the authors as possibly due to missing generator-level filtering in their dataset; if that is the cause, the gap between physics-based and learned mitigation at extreme pileup may be smaller than reported. This is the kind of detail a skeptical reader should check.
  • The architecture is described as generalizable to any domain with dense local signal clusters in global noise (point clouds, anomalous communities), but the paper states this without testing it; a cheap transfer experiment on a non-physics benchmark would indicate whether the claim is specific to collider geometry or general.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes PhyGHT, a hybrid architecture for pileup mitigation at the HL-LHC. The model combines distance-aware graph attention for local geometry, a transformer encoder for global event context, a learnable pileup suppression gate (PSG) that soft-masks tracks, and a hypergraph attention block that aggregates purified tracks to per-jet predictions of energy and mass correction factors. The authors introduce a public simulated t-tbar dataset at <mu>=60 and <mu>=200, benchmark against several baselines including PUPPI, ParticleNet and PUMINet, report improved R2 values, present ablations, and demonstrate a top-quark mass reconstruction as a downstream physics task. The core technical derivation is coherent, and the released dataset/code are a concrete contribution.

Significance. If the reported results are statistically robust, the architecture and the public particle-level dataset would be a useful stepping stone for ML-based pileup mitigation at the HL-LHC. The design idea of making the pileup filter explicit and interpretable via the PSG, while keeping a hierarchical track-to-jet aggregation, is well motivated and goes beyond generic graph/transformer baselines. The paper also contains a useful ablation analysis showing the relative contribution of each block. However, the evaluation currently rests on single-seed runs, unclear train/validation/test discipline for hyperparameter selection, and several oracle-in-the-loop elements in the physics demonstrations. These issues need to be resolved before the central claim of consistent superiority over baselines can be accepted.

major comments (4)
  1. [Section 4.1.3, Tables 4-5] Section 4.1.3 states that k=8 and lambda_aux=0.1 were chosen 'based on the hyperparameter analysis in Section 4.3.2.' Section 4.3.2 (Tables 4 and 5) reports R2 values but does not state which split is used. If these tables are evaluated on the same 10% test split that produces Table 1, then the test set has been used for model selection, and the Table 1 R2 values are optimistically biased. Moreover, the baselines (PUMINet, ParticleNet, etc.) are not reported to have been tuned on the same basis, so the comparison is not on a level field. Please clarify the split used for hyperparameter selection and, if it is the test split, re-run selection on the validation split or use nested cross-validation.
  2. [Appendix B and Table 1] Appendix B reports 'a random seed of 42 for reproducibility' and all experiments appear to be single-seed runs. Table 1 shows PhyGHT's advantage over PUMINet at <mu>=200 is 0.932-0.926=0.006 in energy R2 and 0.836-0.805=0.031 in mass R2. Without seed-to-seed variance (or any uncertainty estimate), it is not possible to assess whether this difference is significant. This directly affects the claim of 'consistently outperforms all baselines' (Section 4.2.1). Please run at least 5 seeds for PhyGHT and the closest baselines and report mean +/- std (or confidence intervals).
  3. [Appendix D and Appendix E] Appendix D: the top-quark mass reconstruction selects W1, W2, and B jets by tracing the parton shower history and using truth labels. Appendix A.8/E: the PUPPI baseline constructs its chi2 metric 'using truth labels to identify pileup contributions.' These oracle uses of generator-level information mean that (i) the downstream physics study does not demonstrate how PhyGHT would perform with a realistic candidate-selection procedure, and (ii) the PUPPI comparison is not a comparison to standard PUPPI as deployed in ATLAS/CMS. Please either replace truth-based selection with reconstructed-object selection, or explicitly frame these as oracle-assisted feasibility checks and adjust the wording in Sections 4.4.1 and 5.
  4. [Section 2.1 and Section 7] Section 2.1 describes a particle-level simulation: neutral particles and charged tracks with pT<400 MeV are removed, vertices are Gaussian-smeared, and no calorimeter response, reconstruction efficiency, fake tracks, or other detector effects are included. Section 7 states 'Our next step is to integrate the model into the ATLAS software framework to process real collision data with full detector effects.' This contradicts the claim in Section 2 that the dataset 'closely mimic[s] the real-world detector data,' and the claim in Section 4.4.1 that the model 'can be used in real world physics analysis.' Please temper these claims to a particle-level benchmark and state the simulation limitations more prominently.
minor comments (5)
  1. [Section 3.7.2] Typo: 'auxillary' should be 'auxiliary'.
  2. [Section 4.1.2 and Table 1] PUMINet and ParticleNet are described as 'state-of-the-art baselines from the ATLAS and CMS experiments,' but PUMINet is not an ATLAS/CMS production algorithm; rephrase to 'baselines proposed in the literature.'
  3. [Appendix E] Minor language issues: 'Heavyside' should be 'Heaviside'; 'derived an R2 score' is acceptable but consider 'computed an R2 score.'
  4. [Figure 2 caption] Inconsistent spacing in '⟨𝜇⟩=60(left)' and '⟨𝜇⟩=200(right)'; add spaces for readability.
  5. [General] The manuscript ends with 'Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009' — this is a template artifact and should be removed.

Circularity Check

1 steps flagged

Central regression is a held-out supervised evaluation and is not circular; the PUPPI baseline is oracle-fitted, and supporting physics demonstrations use truth labels, creating partial evaluation circularity.

specific steps
  1. fitted input called prediction [Section 4.1.2 / Appendix A.8 and Appendix E, Eqs. (23)-(24)]
    "Using truth labels to identify pileup contributions, we construct a χ2 metric from the median and RMS of pileup α values, which is then converted to PUPPI weights via the cumulative χ2 distribution function. These weights are applied to reweight jet constituents, allowing calculation of predicted energy and mass fractions that can be validated against true pileup labels through R2 scores and ROC curves."

    The PUPPI baseline's weights are computed directly from the true pileup labels (the median/RMS of α are estimated from tracks known to be pileup), and the resulting 'predicted' energy/mass fractions are then scored against those same true labels. This is a fitted parameter (the PUPPI weight) being evaluated on the data used to fit it, so the reported PUPPI R2 in Table 1 is not an independent prediction. The comparison is therefore partially circular, though the oracle advantage is against PhyGHT, so it does not inflate the main result.

full rationale

The central claim—PhyGHT's R2 for energy and mass correction factors—rests on a conventional supervised regression with an 80/10/10 split (Section 4.1.1) and a regression loss (Eq. 20) evaluated on a held-out test set. That evaluation is self-contained and not circular: the model does not receive the test labels during training, and no equation reduces the predicted correction factors to the ground-truth ratios by construction. The auxiliary PSG loss (Eq. 21) does share the same per-track vertex labels that define the regression targets (Eq. 1), which reduces the independence of the 'physics-guided' narrative, but the final jet-level predictions are produced by separate learned heads on held-out jets, so this is a label-coupling concern rather than a formal circularity. The clearest circular step is the PUPPI baseline: Appendix E explicitly uses truth labels to build the χ2/PUPPI weights and then validates those weighted predictions against true pileup labels, making the Table 1 PUPPI row an oracle-fitted quantity scored on its fitting labels. Appendix D's top-quark mass reconstruction also uses truth labels to select the W and b jets, so the physics demonstration is not a closed real-data analysis; this is oracle leakage rather than a derivation-level circularity. Hyperparameter choices (k=8, λ_aux=0.1) are reported as selected via Section 4.3.2 without stating whether the tuning curves are on validation or test, and only one seed is used; these are statistical-validity risks, not circularity. Overall, the paper's central result is not forced by definition, but the oracle-augmented baseline and truth-labeled demonstration merit a modest partial-circularity score.

Axiom & Free-Parameter Ledger

11 free parameters · 6 axioms · 0 invented entities

The empirical claim rests on the simulation and on the chosen hyperparameters rather than on a physics derivation. The model weights are learned from data (not listed individually), but the architectural and dataset choices listed above are hand-set and materially affect the reported R2. No new physical entities are postulated.

free parameters (11)
  • Local k-NN neighborhood size k = 8
    Chosen via sweep reported in Table 4; affects the local DA-GAT receptive field and hence energy/mass R2. Selection appears to use test-set R2.
  • Auxiliary loss weight lambda_aux = 0.1
    Chosen via sweep in Table 5; balances regression and track-classification losses. Selected on test-set performance.
  • Hidden dimension D = 128
    Common hyperparameter shared by all models; not swept.
  • Number of attention heads = 4
    Shared hyperparameter for transformer and attention layers; not swept.
  • Network depth L = 3
    Number of layers; not swept.
  • Dropout rate = 0.1
    Regularization; not swept.
  • Learning rate = 3e-4
    AdamW setting; not swept.
  • Batch size = 16 (<mu>=60), 4 (<mu>=200)
    Memory-driven choice.
  • Jet cone radius R = 0.4
    Dataset construction parameter for anti-kt clustering; standard choice but hand-set.
  • Track acceptance cuts = pT>400 MeV; remove neutrals
    Dataset construction; removes detector-level effects and defines the input representation.
  • Vertex smearing widths = sigma_xy=0.3 mm, sigma_z=50 mm
    Gaussian smearing of primary/pileup vertices; hand-set simulation parameter.
axioms (6)
  • domain assumption The Pythia/MadGraph particle-level simulation with acceptance cuts is a faithful proxy for HL-LHC detector data.
    Used throughout Section 2; limitations state full detector simulation is future work, so central claim depends on this proxy being adequate.
  • domain assumption Signal tracks are collimated in (eta,phi) while pileup is uniformly distributed, so local geometric attention can separate them.
    Motivates DA-GAT in Section 3.2; not proven for <mu>=200.
  • domain assumption Track origin truth labels (y_label) are available for training and evaluation.
    Defines the auxiliary loss (Eq. 21) and the regression targets (Eq. 1); in real data such labels must be inferred.
  • standard math Standard ML building blocks (attention, GELU, LayerNorm, message passing) work as intended.
    Used throughout; no formal proof required for empirical claim.
  • domain assumption anti-kt jets with R=0.4 and pT>25 GeV are the correct object definition for the downstream top-mass study.
    Dataset construction; choice affects the benchmark but is not argued.
  • domain assumption The correction-factor approach (multiply raw jet energy/mass by predicted fractions) can recover hard-scatter observables.
    Equation 2; assumes the model's predicted fraction applied to raw jet restores HS quantities.

pith-pipeline@v1.3.0-alltime-deepseek · 16513 in / 16964 out tokens · 154832 ms · 2026-08-02T21:19:20.070696+00:00 · methodology

0 comments
read the original abstract

The High-Luminosity Large Hadron Collider (HL-LHC) at CERN will produce unprecedented datasets capable of revealing fundamental properties of the universe. However, realizing its discovery potential faces a significant challenge: extracting small signal fractions from overwhelming backgrounds dominated by approximately 200 simultaneous pileup collisions. This extreme noise severely distorts the physical observables required for accurate reconstruction. To address this, we introduce the Physics-Guided Hypergraph Transformer (PhyGHT), a hybrid architecture that combines distance-aware local graph attention with global self-attention to mirror the physical topology of particle showers formed in proton-proton collisions. Crucially, we integrate a Pileup Suppression Gate (PSG), an interpretable, physics-constrained mechanism that explicitly learns to filter soft noise prior to hypergraph aggregation. To validate our approach, we release a novel simulated dataset of top-quark pair production to model extreme pileup conditions. PhyGHT outperforms state-of-the-art baselines from the ATLAS and CMS experiments in predicting the signal's energy and mass correction factors. By accurately reconstructing the top quark's invariant mass, we demonstrate how machine learning innovation and interdisciplinary collaboration can directly advance scientific discovery at the frontiers of experimental physics and enhance the HL-LHC's discovery potential. The dataset and code are available at https://github.com/rAIson-Lab/PhyGHT

Figures

Figures reproduced from arXiv: 2602.20475 by Alexander Khanov, Atriya Sen, Flera Rizatdinova, Luke Vaughan, Mohammed Rakib, Shivang Patel.

Figure 1
Figure 1. Figure 1: Signal Collision (Blue) covered in Pileup Collisions (Red) in the HL-LHC Environment [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A Phoenix Event Display[19] depicting an event in the ATLAS inner tracking system in the HL-LHC conditions at ⟨𝜇⟩ = 60 (left) and ⟨𝜇⟩ = 200 (right) for signal (blue) and background (red) particles with 𝑝𝑇 > 1.0 GeV. signal jets become significantly distorted compared to their true values, which hinders physics analysis. The task of pileup mitiga￾tion is to suppress the pileup background and restore the phy… view at source ↗
Figure 3
Figure 3. Figure 3: Overview of PhyGHT. It takes a heterogeneous graph of Tracks and Jets as input. It processes tracks through the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: 1D distribution and 2D correlation plots for Energy ( [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of Energy and Mass Resolutions of PhyGHT with baselines at [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Mass Reconstruction after PhyGHT correction [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Validation Plots for implementation of the PUPPI [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Probing SMEFT Operators through $t\bar{t}t\bar{t}$ Production with Hyper-Graph Neural Networks at the LHC

    hep-ph 2026-05 unverdicted novelty 5.0

    A hyper-graph neural network improves discrimination of four-top production at 13 TeV, raising expected significance from 5.13 to 9.11 and enabling projected 95% CL limits on five dimension-six SMEFT Wilson coefficien...

Reference graph

Works this paper leans on

34 extracted references · 3 canonical work pages · cited by 1 Pith paper

  1. [1]

    Tagging and suppression of pileup jets with the ATLASdetector

    2014. Tagging and suppression of pileup jets with the ATLASdetector. Technical Report. CERN, Geneva. https://cds.cern.ch/record/1700870

    arXiv 2014

  2. [2]

    Graph Neural Network Jet Flavour Tagging with the ATLAS Detector

    2022. Graph Neural Network Jet Flavour Tagging with the ATLAS Detector. Technical Report. CERN, Geneva. https://cds.cern.ch/ record/2811135 All figures including auxiliary figures are available at https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/PUBNOTES/ATL-PHYS- PUB-2022-027

    arXiv 2022

  3. [3]

    O Aberle, C Adorisio, A Adraktas, M Ady, J Albertone, L Alberty, M Alcaide Leon, A Alekou, D Alesini, B Almeida Ferreira, et al . 2020. High-luminosity large hadron collider (HL-LHC): Technical design report. (2020)

    2020

  4. [4]

    Johan Alwall, Rikkert Frederix, Stefano Frixione, Valentin Hirschi, Fabio Maltoni, Olivier Mattelaer, Hua-Sheng Shao, Tim Stelzer, Paolo Torrielli, and Marco Zaro

  5. [5]

    Vaswani Ashish. 2017. Attention is all you need.Advances in neural information processing systems 30 (2017), I

    2017

  6. [6]

    Song Bai, Feihu Zhang, and Philip HS Torr. 2021. Hypergraph convolution and hypergraph attention. Pattern Recognition 110 (2021), 107637

    2021

  7. [7]

    Richard D Ball, Valerio Bertone, Stefano Carrazza, Christopher S Deans, Luigi Del Debbio, Stefano Forte, Alberto Guffanti, Nathan P Hartland, José I Latorre, Juan Rojo, et al. 2013. Parton distributions with LHC data. Nuclear Physics B 867, 2 (2013), 244–289

    2013

  8. [8]

    Daniele Bertolini, Philip Harris, Matthew Low, and Nhan Tran. 2014. Pileup per particle identification. Journal of High Energy Physics (Online) 2014 (10 2014). doi:10.1007/JHEP10(2014)059

    doi:10.1007/jhep10(2014)059 2014

  9. [9]

    Christian Bierlich, Smita Chakraborty, Nishita Desai, Leif Gellersen, Ilkka Hele- nius, Philip Ilten, Leif Lönnblad, Stephen Mrenna, Stefan Prestel, Christian Tobias Preuss, et al. 2022. A comprehensive guide to the physics and usage of PYTHIA 8.3. SciPost Physics Codebases (2022), 008

    2022

  10. [10]

    Matteo Cacciari, Gavin P Salam, and Gregory Soyez. 2008. The anti-kt jet clus- tering algorithm. Journal of High Energy Physics 2008, 04 (2008), 063

    2008

  11. [11]

    Matteo Cacciari, Gavin P Salam, and Gregory Soyez. 2012. FastJet user manual: (for version 3.0. 2). The European Physical Journal C 72, 3 (2012), 1896

    2012

  12. [12]

    Matteo Cacciari, Gavin P Salam, and Gregory Soyez. 2015. SoftKiller, a particle- level pileup removal method. The European Physical Journal C 75, 2 (2015), 59

    2015

  13. [13]

    Carlson, Stephen T

    Benjamin T. Carlson, Stephen T. Roche, Michael Hemmett, and Tae Min Hong

  14. [14]

    ATLAS Collaboration. 2026. Expected Transformer-Network-based jet flavour tag- ging performance with the ATLAS Inner Tracker Detector at the High-Luminosity LHC. in preparation

    2026

  15. [15]

    ATLAS collaboration et al. 2021. Expected tracking and related performance with the updated ATLAS Inner Tracker layout at the High-Luminosity LHC. Techni- cal Report. LHC/ATLAS Experiment

    2021

  16. [16]

    The ATLAS Collaboration. 2014. ATLAS Pythia 8 tunes to 7 TeVdata. Technical Report ATL-PHYS-PUB-2014-021. CERN. https://cds.cern.ch/record/1966419

    arXiv 2014

  17. [17]

    The ATLAS Collaboration and G Aad et al. 2008. The ATLAS Experiment at the CERN Large Hadron Collider. Journal of Instrumentation 3, 08 (aug 2008), S08003. doi:10.1088/1748-0221/3/08/S08003

    doi:10.1088/1748-0221/3/08/s08003 2008

  18. [18]

    The CMS Collaboration and S Chatrchyan et al. 2008. The CMS experiment at the CERN LHC. Journal of Instrumentation 3, 08 (aug 2008), S08004. doi:10.1088/1748- 0221/3/08/S08004

    doi:10.1088/1748- 2008

  19. [19]

    Fawad Ali et al. 2024. HSF/phoenix: v3.0.3. doi:10.5281/zenodo.14203030

    work page doi:10.5281/zenodo.14203030 2024

  20. [20]

    Yifan Feng, Haoxuan You, Zizhao Zhang, Rongrong Ji, and Yue Gao. 2019. Hy- pergraph neural networks. In Proceedings of the AAAI conference on artificial intelligence, Vol. 33. 3558–3565

    2019

  21. [21]

    Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. Advances in neural information processing systems 30 (2017)

    2017

  22. [22]

    Nilotpal Kakati, Etienne Dreyer, Anna Ivina, Francesco Armando Di Bello, Lukas Heinrich, Marumi Kado, and Eilam Gross. 2025. HGPflow: Extending Hypergraph Particle Flow to Collider Event Reconstruction. arXiv:2410.23236 [hep-ex] https: //arxiv.org/abs/2410.23236

    Pith/arXiv arXiv 2025

  23. [23]

    Gregor Kasieczka, Benjamin Nachman, David Shih, Oz Amram, Anders An- dreassen, Kees Benkendorfer, Blaz Bortolato, Gustaaf Brooijmans, Florencia Canelli, Jack H Collins, et al . 2021. The LHC Olympics 2020 a community challenge for anomaly detection in high energy physics. Reports on progress in physics 84, 12 (2021), 124201

    2021

  24. [24]

    Komiske, Eric M

    Patrick T. Komiske, Eric M. Metodiev, Benjamin Nachman, and Matthew D. Schwartz. 2017. Pileup Mitigation with Machine Learning (PUMML). Journal of High Energy Physics 2017, 12 (Dec. 2017). doi:10.1007/jhep12(2017)051

    work page doi:10.1007/jhep12(2017)051 2017

  25. [25]

    B Maier, S M Narayanan, G de Castro, M Goncharov, Ch Paus, and M Schott. 2022. Pile-up mitigation using attention. Machine Learning: Science and Technology 3, 2 (June 2022), 025012. doi:10.1088/2632-2153/ac7198

    work page doi:10.1088/2632-2153/ac7198 2022

  26. [26]

    Jesus Arjona Martinez, Olmo Cerri, Maurizio Pierini, Maria Spiropulu, and Jean- Roch Vlimant. 2019. Pileup mitigation at the Large Hadron Collider with Graph Neural Networks. arXiv:1810.07988 [hep-ph] https://arxiv.org/abs/1810.07988

    Pith/arXiv arXiv 2019

  27. [27]

    Mikuni and F

    V. Mikuni and F. Canelli. 2020. ABCNet: an attention-based method for particle tagging. The European Physical Journal Plus 135, 6 (June 2020). doi:10.1140/ epjp/s13360-020-00497-3

    2020

  28. [28]

    Veličković Petar, Cucurull Guillem, Casanova Arantxa, Romero Adriana, Lio Pietro, and B Yoshua. 2018. Graph attention networks. InInternational conference on learning representations, Vol. 8

    2018

  29. [29]

    Huilin Qu and Loukas Gouskos. 2020. Jet tagging via particle clouds. Physical Review D 101, 5 (2020), 056019

    2020

  30. [30]

    Huilin Qu, Congqiao Li, and Sitian Qian. 2024. Particle Transformer for Jet Tagging. arXiv:2202.03772 [hep-ph] https://arxiv.org/abs/2202.03772

    Pith/arXiv arXiv 2024

  31. [31]

    David Rousseau, Sabrina Amrouche, Paolo Calafiura, Victor Estrade, Steven Farrell, Cecile Germain, Vladimir Gligorov, Tobias Golling, Heather Gray, Isabelle Guyon, et al. 2018. The TrackML Particle Tracking Challenge. (2018)

    2018

  32. [32]

    Luke Vaughan, Mohammed Rakib, Shivang Patel, Flera Rizatdinova, Alexander Khanov, and Arunkumar Bagavathi. 2025. PileUp Mitigation at the HL-LHC Using Attention for Event-Wide Context. In Pacific-Asia Conference on Knowledge Discovery and Data Mining. Springer, 342–353. KDD ’26, August 09–13, 2026, Jeju, Korea Rakib et al. Appendix A Baseline Descriptions...

    2025

  33. [2014]

    Journal of High Energy Physics 2014 (2014)

    The automated computation of tree-level and next-to-leading order differ- ential cross sections, and their matching to parton shower simulations. Journal of High Energy Physics 2014 (2014). https://api.semanticscholar.org/CorpusID: 256012920

    2014

  34. [2025]

    arXiv:2507.16686 [hep-ph] https://arxiv.org/abs/2507.16686

    Ring-based ML calibration with in situ pileup correction for real-time jet triggers. arXiv:2507.16686 [hep-ph] https://arxiv.org/abs/2507.16686

    Pith/arXiv arXiv