Pretrained Event Classification Model for High Energy Physics Analysis

Benjamin Ryan Roberts; Haichen Wang; Joshua Ho; Shuo Han

arxiv: 2412.10665 · v2 · submitted 2024-12-14 · ✦ hep-ph · cs.LG

Pretrained Event Classification Model for High Energy Physics Analysis

Joshua Ho , Benjamin Ryan Roberts , Shuo Han , Haichen Wang This is my paper

Pith reviewed 2026-05-23 07:20 UTC · model grok-4.3

classification ✦ hep-ph cs.LG

keywords pretrained modelevent classificationgraph neural networkhigh energy physicsfine-tuningcentered kernel alignmentfoundation modelproton-proton collisions

0 comments

The pith

A graph neural network pretrained on 120 million collision events improves fine-tuned classification accuracy and efficiency on new high-energy physics tasks, especially with limited data.

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

The paper introduces a foundation model for event classification in high-energy physics built on a graph neural network and pretrained on 120 million simulated proton-proton collision events across 12 processes using multiclass and multilabel tasks. Evaluation across seven tasks, including unseen physics processes and ATLAS Open Data from different simulation frameworks, shows that fine-tuning yields gains in accuracy and computational efficiency particularly when training data is scarce. A centered kernel alignment analysis indicates that encoder-stage representations remain similar to the baseline while intermediate graph processing layers diverge substantially, pointing to the development of task-specific message-passing pathways.

Core claim

The model learns general representations through pretraining on large-scale simulated data and, upon fine-tuning, achieves superior performance on downstream classification tasks not encountered during pretraining while preserving encoder representations and altering intermediate graph layers as revealed by centered kernel alignment.

What carries the argument

The pretrained Graph Neural Network with encoder stages and intermediate graph processing layers, analyzed via Centered Kernel Alignment to track representational changes during fine-tuning.

If this is right

Classification accuracy improves on tasks involving new physics processes absent from pretraining.
Performance and efficiency gains are largest when the amount of labeled training data is small.
The model generalizes across simulation frameworks from fast Delphes to full ATLAS detector simulation.
Encoder representations stay largely unchanged while message-passing pathways adapt during fine-tuning.

Where Pith is reading between the lines

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

Pretraining could lower the total compute required for developing classifiers on future datasets.
The approach might extend to other high-energy physics tasks such as regression or anomaly detection.
Real-data validation would need explicit tests against unmodeled systematic uncertainties.

Load-bearing premise

Gains measured on simulated test sets and ATLAS Open Data will carry over to real experimental data that includes detector effects, backgrounds, and systematic uncertainties not fully present in the simulations.

What would settle it

Measuring the fine-tuned model's classification performance directly on recorded LHC collision data and comparing it to results on the simulated test sets used in the paper.

Figures

Figures reproduced from arXiv: 2412.10665 by Benjamin Ryan Roberts, Haichen Wang, Joshua Ho, Shuo Han.

read the original abstract

We introduce a foundation model for event classification in high-energy physics, built on a Graph Neural Network architecture and trained on 120 million simulated proton-proton collision events spanning 12 distinct physics processes. The model is pretrained to learn a general and robust representation of collision data using challenging multiclass and multilabel classification tasks. Its performance is evaluated across seven event classification tasks, which include new physics processes not encountered during pretraining as well as ATLAS Open Data to demonstrate generalizability across different simulation frameworks, from Delphes fast simulation to full ATLAS detector simulation. Fine-tuning the pretrained model significantly improves classification performance, particularly in scenarios with limited training data, demonstrating gains in both accuracy and computational efficiency. To investigate the underlying mechanisms behind these performance improvements, we employ a representational similarity evaluation framework based on Centered Kernel Alignment. This analysis reveals that encoder-stage representations of the fine-tuned model remain similar to those of the baseline, while intermediate graph processing layers diverge substantially, indicating that fine-tuning preserves general-purpose encoders while developing fundamentally different message-passing pathways to arrive at superior task performance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper shows concrete gains from pretraining a GNN on 120M events and uses CKA to track what changes in fine-tuning, but lacks numbers to judge the effect size.

read the letter

The main takeaway is that pretraining a Graph Neural Network on 120 million simulated proton-proton events across 12 processes improves classification on seven downstream tasks, including new physics channels and ATLAS Open Data from different simulators, with the biggest help when training data is limited. The CKA analysis indicates that encoder representations stay close to the baseline while intermediate message-passing layers shift more, which is a reasonable way to probe the source of the gains.

Referee Report

2 major / 1 minor

Summary. The paper introduces a GNN-based foundation model pretrained on 120 million simulated proton-proton collision events spanning 12 physics processes for multiclass and multilabel classification tasks. It evaluates fine-tuning on seven downstream event classification tasks (including unseen new physics processes and ATLAS Open Data across Delphes and full ATLAS simulation frameworks), reports gains in accuracy and efficiency especially in low-data regimes, and applies Centered Kernel Alignment (CKA) to show that encoder-stage representations remain similar to the baseline while intermediate graph-processing layers diverge substantially.

Significance. If the reported empirical gains and CKA observations hold under full experimental scrutiny, the work would demonstrate a practical route to transfer learning for HEP event classification, with potential value for analyses constrained by limited labeled data. The scale of pretraining (120M events) and the cross-simulation evaluation constitute concrete strengths; the CKA analysis supplies a mechanistic probe of what fine-tuning modifies. These elements could inform future foundation-model efforts in the field provided the quantitative improvements are robustly documented.

major comments (2)

[Abstract] Abstract and results sections: the central claim that fine-tuning 'significantly improves classification performance' is stated without accompanying numerical values (accuracy, AUC, or F1), baseline comparisons, error bars, or statistical significance tests for the seven tasks. This absence prevents verification of the magnitude and reliability of the reported gains.
[CKA analysis] CKA analysis paragraph: the statement that 'intermediate graph processing layers diverge substantially' is presented without the actual CKA similarity matrices, layer indices, or quantitative thresholds used to define 'similar' versus 'diverge.' The section must supply these values and the precise definition of the CKA metric employed.

minor comments (1)

[Evaluation] The description of the seven evaluation tasks would benefit from an explicit table listing the processes, training-set sizes, and simulation frameworks for each task.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to improve clarity and completeness.

read point-by-point responses

Referee: [Abstract] Abstract and results sections: the central claim that fine-tuning 'significantly improves classification performance' is stated without accompanying numerical values (accuracy, AUC, or F1), baseline comparisons, error bars, or statistical significance tests for the seven tasks. This absence prevents verification of the magnitude and reliability of the reported gains.

Authors: The results section presents the performance metrics, baseline comparisons, and error bars for all seven tasks in tables and figures. To address the concern and make the abstract self-contained, we will add explicit numerical values for key accuracy and AUC improvements (particularly in low-data regimes), along with references to the statistical comparisons, in the revised abstract. revision: yes
Referee: [CKA analysis] CKA analysis paragraph: the statement that 'intermediate graph processing layers diverge substantially' is presented without the actual CKA similarity matrices, layer indices, or quantitative thresholds used to define 'similar' versus 'diverge.' The section must supply these values and the precise definition of the CKA metric employed.

Authors: The manuscript includes figures displaying the CKA similarity matrices. We will revise the text to explicitly list the layer indices, report the quantitative CKA values, provide the standard definition of the CKA metric as used in the analysis, and state the thresholds applied to classify representations as similar or substantially divergent. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper describes an empirical machine-learning study: pretraining a GNN on 120M simulated events, fine-tuning on seven classification tasks, and measuring accuracy/CKA similarity. All reported results are direct outcomes of training runs and post-hoc similarity metrics on held-out simulated data. No derivation chain, first-principles claim, fitted parameter renamed as prediction, or self-citation that bears the central result is present. The work is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work relies on standard machine-learning assumptions about simulated data fidelity and the validity of CKA as a similarity measure; no new free parameters, axioms, or invented entities are introduced beyond conventional GNN training practices.

axioms (1)

domain assumption Simulated collision events from Delphes and ATLAS frameworks sufficiently approximate real detector data for pretraining and evaluation purposes
The model is pretrained exclusively on simulation and evaluated partly on ATLAS Open Data from different simulation chains.

pith-pipeline@v0.9.0 · 5716 in / 1360 out tokens · 29586 ms · 2026-05-23T07:20:36.264650+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages · 16 internal anchors

[1]

Multi-class Classification For Monte Carlo simulated events, the underlying physics process that generated each event is known pre- cisely, providing natural labels for supervised learning. However, the challenge lies in the complexity of collision events: different physics processes can produce similar kinematics and event topologies, particularly in cer...

work page
[2]

For discrete properties like particle presence in specific kinematic re- gions, we employ classification labels with binary cross- entropy loss

Multi-label Classification This approach combines both classification and regres- sion tasks to characterize collision events. For discrete properties like particle presence in specific kinematic re- gions, we employ classification labels with binary cross- entropy loss. For continuous quantities like particle mul- tiplicities, we use regression labels wi...

work page
[3]

head start,

Pretraining During pre-training, the initial learning rate is 10 −4, and the learning rate decays by 1% each epoch following the power law function LR(x) = 10−4 · (0.99)x, where x is the number of epochs. Both pre-trained models reach a plateau in loss by epoch 50, at which point the training is stopped. D. Fine-tuning Methodology For downstream tasks, we...

work page
[4]

GPT-4 Technical Report

OpenAI, J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, R. Avila, I. Babuschkin, S. Bal- aji, V. Balcom, P. Baltescu, H. Bao, M. Bavarian, J. Belgum, I. Bello, J. Berdine, G. Bernadett-Shapiro, C. Berner, L. Bogdonoff, O. Boiko, M. Boyd, A.-L. Brak- man, G. Brockman, T. Brooks, M. Bru...

work page internal anchor Pith review Pith/arXiv arXiv 2024
[5]

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

J. Devlin, M. Chang, K. Lee, and K. Toutanova, BERT: pre-training of deep bidirectional transformers for lan- guage understanding, CoRR abs/1810.04805 (2018), 1810.04805

work page internal anchor Pith review Pith/arXiv arXiv 2018
[6]

High-Resolution Image Synthesis with Latent Diffusion Models

R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, High-resolution image synthesis with la- tent diffusion models, CoRR abs/2112.10752 (2021), 2112.10752

work page internal anchor Pith review Pith/arXiv arXiv 2021
[7]

SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis

D. Podell, Z. English, K. Lacey, A. Blattmann, T. Dock- horn, J. M¨ uller, J. Penna, and R. Rombach, Sdxl: Im- proving latent diffusion models for high-resolution image synthesis (2023), arXiv:2307.01952 [cs.CV]

work page internal anchor Pith review Pith/arXiv arXiv 2023
[8]

Jumper, R

J. Jumper, R. Evans, A. Pritzel, et al. , Highly accurate protein structure prediction with alphafold, Nature 596, 583 (2021)

work page 2021
[9]

How transferable are features in deep neural networks?

J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, How transferable are features in deep neural networks?, CoRR abs/1411.1792 (2014), 1411.1792

work page internal anchor Pith review Pith/arXiv arXiv 2014
[10]

A. J. Wildridge, J. P. Rodgers, E. M. Colbert, Y. yao, A. W. Jung, and M. Liu, Bumblebee: Foundation model for particle physics discovery (2024), arXiv:2412.07867 [hep-ex]

work page arXiv 2024
[11]

Katel, H

S. Katel, H. Li, Z. Zhao, R. Kansal, F. Mokhtar, and J. Duarte, Learning symmetry-independent jet represen- tations via jet-based joint embedding predictive architec- ture (2024), arXiv:2412.05333 [hep-ph]

work page arXiv 2024
[12]

Golling, L

T. Golling, L. Heinrich, M. Kagan, S. Klein, M. Leigh, M. Osadchy, and J. A. Raine, Masked particle model- ing on sets: Towards self-supervised high energy physics foundation models (2024), arXiv:2401.13537 [hep-ph]

work page arXiv 2024
[13]

Mikuni and B

V. Mikuni and B. Nachman, Omnilearn: A method to simultaneously facilitate all jet physics tasks (2024), arXiv:2404.16091 [hep-ph]

work page arXiv 2024
[14]

Harris, M

P. Harris, M. Kagan, J. Krupa, B. Maier, and N. Wood- ward, Re-simulation-based self-supervised learning for pre-training foundation models (2024), arXiv:2403.07066 [hep-ph]

work page arXiv 2024
[15]

J. Birk, A. Hallin, and G. Kasieczka, Omnijet-α: the first cross-task foundation model for particle physics, Machine Learning: Science and Technology 5, 035031 (2024)

work page 2024
[16]

M. Vigl, N. Hartman, and L. Heinrich, Finetuning foun- dation models for joint analysis optimization (2024), arXiv:2401.13536 [hep-ex]

work page arXiv 2024
[17]

J. Y. Araz, V. Mikuni, F. Ringer, N. Sato, F. T. Acosta, and R. Whitehill, Point cloud-based diffusion models for the electron-ion collider (2024), arXiv:2410.22421 [hep- ph]

work page arXiv 2024
[18]

J. Liu, A. Ghosh, D. Smith, P. Baldi, and D. Whiteson, Generalizing to new geometries with geometry-aware au- toregressive models (gaams) for fast calorimeter simula- tion, Journal of Instrumentation 18 (11), P11003

work page
[19]

Hashemi, N

B. Hashemi, N. Hartmann, S. Sharifzadeh, J. Kahn, and T. Kuhr, Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self- supervised relational reasoning, Nature Communications 15, 10.1038/s41467-024-49104-4 (2024)

work page doi:10.1038/s41467-024-49104-4 2024
[20]

Huang, Y

A. Huang, Y. Melkani, P. Calafiura, A. Lazar, D. T. Mur- nane, M.-T. Pham, and X. Ju, A language model for par- ticle tracking (2024), arXiv:2402.10239 [hep-ph]

work page arXiv 2024
[21]

Zhang, Y

Z. Zhang, Y. Zhang, H. Yao, J. Luo, R. Zhao, B. Huang, J. Zhao, Y. Liao, K. Li, L. Zhao, J. Cao, F. Qi, and 9 C. Yuan, Xiwu: A basis flexible and learnable llm for high energy physics (2024), arXiv:2404.08001 [hep-ph]

work page arXiv 2024
[22]

ATLAS Collaboration (ATLAS), Observation of four- top-quark production in the multilepton final state with the ATLAS detector, Eur. Phys. J. C83, 496 (2023), [Er- ratum: Eur.Phys.J.C 84, 156 (2024)], arXiv:2303.15061 [hep-ex]

work page arXiv 2023
[23]

The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations

J. Alwall, R. Frederix, S. Frixione, V. Hirschi, F. Maltoni, O. Mattelaer, H. S. Shao, T. Stelzer, P. Torrielli, and M. Zaro, The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations, JHEP 07, 079, arXiv:1405.0301 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv
[24]

A framework for Higgs characterisation

P. Artoisenet et al., A framework for Higgs characterisa- tion, JHEP 11, 043, arXiv:1306.6464 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv
[25]

Rosiek, Complete set of feynman rules for the minimal supersymmetric extension of the standard model, Phys

J. Rosiek, Complete set of feynman rules for the minimal supersymmetric extension of the standard model, Phys. Rev. D 41, 3464 (1990)

work page 1990
[26]

Allanach, C

B. Allanach, C. Bal´ azs, G. B´ elanger, M. Bernhardt, F. Boudjema, D. Choudhury, K. Desch, U. Ell- wanger, P. Gambino, R. Godbole, T. Goto, J. Guasch, M. Guchait, T. Hahn, S. Heinemeyer, C. Hugonie, T. Hurth, S. Kraml, S. Kreiss, J. Lykken, F. Moort- gat, S. Moretti, S. Pe˜ naranda, T. Plehn, W. Porod, A. Pukhov, P. Richardson, M. Schumacher, L. Sil- ves...

work page 2009
[27]

Degrande, F

C. Degrande, F. Maltoni, J. Wang, and C. Zhang, Au- tomatic computations at next-to-leading order in qcd for top-quark flavor-changing neutral processes, Phys. Rev. D 91, 034024 (2015)

work page 2015
[28]

Durieux, F

G. Durieux, F. Maltoni, and C. Zhang, Global approach to top-quark flavor-changing interactions, Phys. Rev. D 91, 074017 (2015)

work page 2015
[29]

Automatic spin-entangled decays of heavy resonances in Monte Carlo simulations

P. Artoisenet, R. Frederix, O. Mattelaer, and R. Rietkerk, Automatic spin-entangled decays of heavy resonances in Monte Carlo simulations, JHEP03, 015, arXiv:1212.3460 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv
[30]

An Introduction to PYTHIA 8.2

T. Sj¨ ostrand, S. Ask, J. R. Christiansen, R. Corke, N. De- sai, P. Ilten, S. Mrenna, S. Prestel, C. O. Rasmussen, and P. Z. Skands, An introduction to PYTHIA 8.2, Comput. Phys. Commun. 191, 159 (2015), arXiv:1410.3012 [hep- ph]

work page internal anchor Pith review Pith/arXiv arXiv 2015
[31]

Delphes 3: a modular framework for fast simulation of a generic collider experiment.Journal of High Energy Physics, 2014(2):57, 2014

J. de Favereau, C. Delaere, P. Demin, A. Giammanco, V. Lemaˆ ıtre, A. Mertens, and M. Selvaggi (DELPHES 3), DELPHES 3, A modular framework for fast simu- lation of a generic collider experiment, JHEP 02, 057, arXiv:1307.6346 [hep-ex]

work page internal anchor Pith review Pith/arXiv arXiv
[32]

T. A. Collaboration, The atlas experiment at the cern large hadron collider, Journal of Instrumentation 3 (08), S08003

work page
[33]

The anti-k_t jet clustering algorithm

M. Cacciari, G. P. Salam, and G. Soyez, The anti- kt jet clustering algorithm, JHEP 04, 063, arXiv:0802.1189 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv
[34]

Aad et al

G. Aad et al. (ATLAS), ATLAS b-jet identification per- formance and efficiency measurement with t¯t events in pp collisions at √s = 13 TeV, Eur. Phys. J. C 79, 970 (2019), arXiv:1907.05120 [hep-ex]

work page arXiv 2019
[35]

M. Wang, L. Yu, D. Zheng, Q. Gan, Y. Gai, Z. Ye, M. Li, J. Zhou, Q. Huang, C. Ma, Z. Huang, Q. Guo, H. Zhang, H. Lin, J. Zhao, J. Li, A. J. Smola, and Z. Zhang, Deep graph library: Towards efficient and scal- able deep learning on graphs, CoRR abs/1909.01315 (2019), 1909.01315

work page internal anchor Pith review arXiv 1909
[36]

PyTorch: An Imperative Style, High-Performance Deep Learning Library

A. Paszke, S. Gross, F. Massa, A. Lerer, J. Brad- bury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. K¨ opf, E. Z. Yang, Z. De- Vito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, Pytorch: An impera- tive style, high-performance deep learning library, CoRR abs/1912.01703 (2019), 1912.01703

work page internal anchor Pith review Pith/arXiv arXiv 1912
[37]

P. W. Battaglia, J. B. Hamrick, V. Bapst, A. Sanchez- Gonzalez, V. F. Zambaldi, M. Malinowski, A. Tacchetti, D. Raposo, A. Santoro, R. Faulkner, C ¸ . G¨ ul¸ cehre, H. F. Song, A. J. Ballard, J. Gilmer, G. E. Dahl, A. Vaswani, K. R. Allen, C. Nash, V. Langston, C. Dyer, N. Heess, D. Wierstra, P. Kohli, M. M. Botvinick, O. Vinyals, Y. Li, and R. Pascanu, R...

work page internal anchor Pith review Pith/arXiv arXiv 2018
[38]

J. L. Ba, J. R. Kiros, and G. E. Hinton, Layer normal- ization (2016), arXiv:1607.06450 [stat.ML]

work page internal anchor Pith review Pith/arXiv arXiv 2016
[39]

Similarity of Neural Network Representations Revisited

S. Kornblith, M. Norouzi, H. Lee, and G. E. Hinton, Sim- ilarity of neural network representations revisited, CoRR abs/1905.00414 (2019), 1905.00414

work page internal anchor Pith review Pith/arXiv arXiv 1905
[40]

ATLAS Collaboration, Measurement of the properties of Higgs boson production at √s = 13 TeV in the H → γγ channel using 139 fb−1 of pp collision data with the AT- LAS experiment, JHEP 07, 088, arXiv:2207.00348 [hep- ex]

work page arXiv

[1] [1]

Multi-class Classification For Monte Carlo simulated events, the underlying physics process that generated each event is known pre- cisely, providing natural labels for supervised learning. However, the challenge lies in the complexity of collision events: different physics processes can produce similar kinematics and event topologies, particularly in cer...

work page

[2] [2]

For discrete properties like particle presence in specific kinematic re- gions, we employ classification labels with binary cross- entropy loss

Multi-label Classification This approach combines both classification and regres- sion tasks to characterize collision events. For discrete properties like particle presence in specific kinematic re- gions, we employ classification labels with binary cross- entropy loss. For continuous quantities like particle mul- tiplicities, we use regression labels wi...

work page

[3] [3]

head start,

Pretraining During pre-training, the initial learning rate is 10 −4, and the learning rate decays by 1% each epoch following the power law function LR(x) = 10−4 · (0.99)x, where x is the number of epochs. Both pre-trained models reach a plateau in loss by epoch 50, at which point the training is stopped. D. Fine-tuning Methodology For downstream tasks, we...

work page

[4] [4]

GPT-4 Technical Report

OpenAI, J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, R. Avila, I. Babuschkin, S. Bal- aji, V. Balcom, P. Baltescu, H. Bao, M. Bavarian, J. Belgum, I. Bello, J. Berdine, G. Bernadett-Shapiro, C. Berner, L. Bogdonoff, O. Boiko, M. Boyd, A.-L. Brak- man, G. Brockman, T. Brooks, M. Bru...

work page internal anchor Pith review Pith/arXiv arXiv 2024

[5] [5]

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

J. Devlin, M. Chang, K. Lee, and K. Toutanova, BERT: pre-training of deep bidirectional transformers for lan- guage understanding, CoRR abs/1810.04805 (2018), 1810.04805

work page internal anchor Pith review Pith/arXiv arXiv 2018

[6] [6]

High-Resolution Image Synthesis with Latent Diffusion Models

R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, High-resolution image synthesis with la- tent diffusion models, CoRR abs/2112.10752 (2021), 2112.10752

work page internal anchor Pith review Pith/arXiv arXiv 2021

[7] [7]

SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis

D. Podell, Z. English, K. Lacey, A. Blattmann, T. Dock- horn, J. M¨ uller, J. Penna, and R. Rombach, Sdxl: Im- proving latent diffusion models for high-resolution image synthesis (2023), arXiv:2307.01952 [cs.CV]

work page internal anchor Pith review Pith/arXiv arXiv 2023

[8] [8]

Jumper, R

J. Jumper, R. Evans, A. Pritzel, et al. , Highly accurate protein structure prediction with alphafold, Nature 596, 583 (2021)

work page 2021

[9] [9]

How transferable are features in deep neural networks?

J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, How transferable are features in deep neural networks?, CoRR abs/1411.1792 (2014), 1411.1792

work page internal anchor Pith review Pith/arXiv arXiv 2014

[10] [10]

A. J. Wildridge, J. P. Rodgers, E. M. Colbert, Y. yao, A. W. Jung, and M. Liu, Bumblebee: Foundation model for particle physics discovery (2024), arXiv:2412.07867 [hep-ex]

work page arXiv 2024

[11] [11]

Katel, H

S. Katel, H. Li, Z. Zhao, R. Kansal, F. Mokhtar, and J. Duarte, Learning symmetry-independent jet represen- tations via jet-based joint embedding predictive architec- ture (2024), arXiv:2412.05333 [hep-ph]

work page arXiv 2024

[12] [12]

Golling, L

T. Golling, L. Heinrich, M. Kagan, S. Klein, M. Leigh, M. Osadchy, and J. A. Raine, Masked particle model- ing on sets: Towards self-supervised high energy physics foundation models (2024), arXiv:2401.13537 [hep-ph]

work page arXiv 2024

[13] [13]

Mikuni and B

V. Mikuni and B. Nachman, Omnilearn: A method to simultaneously facilitate all jet physics tasks (2024), arXiv:2404.16091 [hep-ph]

work page arXiv 2024

[14] [14]

Harris, M

P. Harris, M. Kagan, J. Krupa, B. Maier, and N. Wood- ward, Re-simulation-based self-supervised learning for pre-training foundation models (2024), arXiv:2403.07066 [hep-ph]

work page arXiv 2024

[15] [15]

J. Birk, A. Hallin, and G. Kasieczka, Omnijet-α: the first cross-task foundation model for particle physics, Machine Learning: Science and Technology 5, 035031 (2024)

work page 2024

[16] [16]

M. Vigl, N. Hartman, and L. Heinrich, Finetuning foun- dation models for joint analysis optimization (2024), arXiv:2401.13536 [hep-ex]

work page arXiv 2024

[17] [17]

J. Y. Araz, V. Mikuni, F. Ringer, N. Sato, F. T. Acosta, and R. Whitehill, Point cloud-based diffusion models for the electron-ion collider (2024), arXiv:2410.22421 [hep- ph]

work page arXiv 2024

[18] [18]

J. Liu, A. Ghosh, D. Smith, P. Baldi, and D. Whiteson, Generalizing to new geometries with geometry-aware au- toregressive models (gaams) for fast calorimeter simula- tion, Journal of Instrumentation 18 (11), P11003

work page

[19] [19]

Hashemi, N

B. Hashemi, N. Hartmann, S. Sharifzadeh, J. Kahn, and T. Kuhr, Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self- supervised relational reasoning, Nature Communications 15, 10.1038/s41467-024-49104-4 (2024)

work page doi:10.1038/s41467-024-49104-4 2024

[20] [20]

Huang, Y

A. Huang, Y. Melkani, P. Calafiura, A. Lazar, D. T. Mur- nane, M.-T. Pham, and X. Ju, A language model for par- ticle tracking (2024), arXiv:2402.10239 [hep-ph]

work page arXiv 2024

[21] [21]

Zhang, Y

Z. Zhang, Y. Zhang, H. Yao, J. Luo, R. Zhao, B. Huang, J. Zhao, Y. Liao, K. Li, L. Zhao, J. Cao, F. Qi, and 9 C. Yuan, Xiwu: A basis flexible and learnable llm for high energy physics (2024), arXiv:2404.08001 [hep-ph]

work page arXiv 2024

[22] [22]

ATLAS Collaboration (ATLAS), Observation of four- top-quark production in the multilepton final state with the ATLAS detector, Eur. Phys. J. C83, 496 (2023), [Er- ratum: Eur.Phys.J.C 84, 156 (2024)], arXiv:2303.15061 [hep-ex]

work page arXiv 2023

[23] [23]

The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations

J. Alwall, R. Frederix, S. Frixione, V. Hirschi, F. Maltoni, O. Mattelaer, H. S. Shao, T. Stelzer, P. Torrielli, and M. Zaro, The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations, JHEP 07, 079, arXiv:1405.0301 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv

[24] [24]

A framework for Higgs characterisation

P. Artoisenet et al., A framework for Higgs characterisa- tion, JHEP 11, 043, arXiv:1306.6464 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv

[25] [25]

Rosiek, Complete set of feynman rules for the minimal supersymmetric extension of the standard model, Phys

J. Rosiek, Complete set of feynman rules for the minimal supersymmetric extension of the standard model, Phys. Rev. D 41, 3464 (1990)

work page 1990

[26] [26]

Allanach, C

B. Allanach, C. Bal´ azs, G. B´ elanger, M. Bernhardt, F. Boudjema, D. Choudhury, K. Desch, U. Ell- wanger, P. Gambino, R. Godbole, T. Goto, J. Guasch, M. Guchait, T. Hahn, S. Heinemeyer, C. Hugonie, T. Hurth, S. Kraml, S. Kreiss, J. Lykken, F. Moort- gat, S. Moretti, S. Pe˜ naranda, T. Plehn, W. Porod, A. Pukhov, P. Richardson, M. Schumacher, L. Sil- ves...

work page 2009

[27] [27]

Degrande, F

C. Degrande, F. Maltoni, J. Wang, and C. Zhang, Au- tomatic computations at next-to-leading order in qcd for top-quark flavor-changing neutral processes, Phys. Rev. D 91, 034024 (2015)

work page 2015

[28] [28]

Durieux, F

G. Durieux, F. Maltoni, and C. Zhang, Global approach to top-quark flavor-changing interactions, Phys. Rev. D 91, 074017 (2015)

work page 2015

[29] [29]

Automatic spin-entangled decays of heavy resonances in Monte Carlo simulations

P. Artoisenet, R. Frederix, O. Mattelaer, and R. Rietkerk, Automatic spin-entangled decays of heavy resonances in Monte Carlo simulations, JHEP03, 015, arXiv:1212.3460 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv

[30] [30]

An Introduction to PYTHIA 8.2

T. Sj¨ ostrand, S. Ask, J. R. Christiansen, R. Corke, N. De- sai, P. Ilten, S. Mrenna, S. Prestel, C. O. Rasmussen, and P. Z. Skands, An introduction to PYTHIA 8.2, Comput. Phys. Commun. 191, 159 (2015), arXiv:1410.3012 [hep- ph]

work page internal anchor Pith review Pith/arXiv arXiv 2015

[31] [31]

Delphes 3: a modular framework for fast simulation of a generic collider experiment.Journal of High Energy Physics, 2014(2):57, 2014

J. de Favereau, C. Delaere, P. Demin, A. Giammanco, V. Lemaˆ ıtre, A. Mertens, and M. Selvaggi (DELPHES 3), DELPHES 3, A modular framework for fast simu- lation of a generic collider experiment, JHEP 02, 057, arXiv:1307.6346 [hep-ex]

work page internal anchor Pith review Pith/arXiv arXiv

[32] [32]

T. A. Collaboration, The atlas experiment at the cern large hadron collider, Journal of Instrumentation 3 (08), S08003

work page

[33] [33]

The anti-k_t jet clustering algorithm

M. Cacciari, G. P. Salam, and G. Soyez, The anti- kt jet clustering algorithm, JHEP 04, 063, arXiv:0802.1189 [hep-ph]

work page internal anchor Pith review Pith/arXiv arXiv

[34] [34]

Aad et al

G. Aad et al. (ATLAS), ATLAS b-jet identification per- formance and efficiency measurement with t¯t events in pp collisions at √s = 13 TeV, Eur. Phys. J. C 79, 970 (2019), arXiv:1907.05120 [hep-ex]

work page arXiv 2019

[35] [35]

M. Wang, L. Yu, D. Zheng, Q. Gan, Y. Gai, Z. Ye, M. Li, J. Zhou, Q. Huang, C. Ma, Z. Huang, Q. Guo, H. Zhang, H. Lin, J. Zhao, J. Li, A. J. Smola, and Z. Zhang, Deep graph library: Towards efficient and scal- able deep learning on graphs, CoRR abs/1909.01315 (2019), 1909.01315

work page internal anchor Pith review arXiv 1909

[36] [36]

PyTorch: An Imperative Style, High-Performance Deep Learning Library

A. Paszke, S. Gross, F. Massa, A. Lerer, J. Brad- bury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. K¨ opf, E. Z. Yang, Z. De- Vito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, Pytorch: An impera- tive style, high-performance deep learning library, CoRR abs/1912.01703 (2019), 1912.01703

work page internal anchor Pith review Pith/arXiv arXiv 1912

[37] [37]

P. W. Battaglia, J. B. Hamrick, V. Bapst, A. Sanchez- Gonzalez, V. F. Zambaldi, M. Malinowski, A. Tacchetti, D. Raposo, A. Santoro, R. Faulkner, C ¸ . G¨ ul¸ cehre, H. F. Song, A. J. Ballard, J. Gilmer, G. E. Dahl, A. Vaswani, K. R. Allen, C. Nash, V. Langston, C. Dyer, N. Heess, D. Wierstra, P. Kohli, M. M. Botvinick, O. Vinyals, Y. Li, and R. Pascanu, R...

work page internal anchor Pith review Pith/arXiv arXiv 2018

[38] [38]

J. L. Ba, J. R. Kiros, and G. E. Hinton, Layer normal- ization (2016), arXiv:1607.06450 [stat.ML]

work page internal anchor Pith review Pith/arXiv arXiv 2016

[39] [39]

Similarity of Neural Network Representations Revisited

S. Kornblith, M. Norouzi, H. Lee, and G. E. Hinton, Sim- ilarity of neural network representations revisited, CoRR abs/1905.00414 (2019), 1905.00414

work page internal anchor Pith review Pith/arXiv arXiv 1905

[40] [40]

ATLAS Collaboration, Measurement of the properties of Higgs boson production at √s = 13 TeV in the H → γγ channel using 139 fb−1 of pp collision data with the AT- LAS experiment, JHEP 07, 088, arXiv:2207.00348 [hep- ex]

work page arXiv