Pith. sign in

REVIEW 5 minor 142 references

Verification of machine learning is essential only when its outputs enter statistical modeling, inference, or hypothesis testing for discovery claims.

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 · grok-4.5

2026-07-14 00:49 UTC pith:FKYLUFS7

load-bearing objection Solid community synthesis that maps when ML verification is load-bearing for discovery claims; useful reference, not a new result.

arxiv 2607.10039 v1 pith:FKYLUFS7 submitted 2026-07-10 physics.data-an cs.LGhep-ph

Are We Ready for AI-Driven Discovery? AI Verification Before the Next Fundamental Physics Breakthrough

Gaia Grosso , Vinicius Mikuni , Lukas Heinrich This is my paper
classification physics.data-an cs.LGhep-ph PACS 07.05.Mh29.85.-c02.50.-r
keywords machine learning verificationstatistical discovery workflowsimulation-based inferenceuncertainty quantificationinductive biasagentic AIfundamental physics
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.

Machine learning now accelerates every stage of fundamental-physics discovery, from triggering and simulation to inference and agentic analysis. The paper argues that reliability for discovery claims does not require perfect models everywhere; it requires verification precisely where ML outputs enter the statistical model used for inference or testing. Elsewhere, imperfect summarization, calibrated surrogates, or exploratory tools are tolerable so long as residual uncertainties are quantified and no unmodeled systematic bias is introduced. The authors also map irreducible limits—unavoidable inductive bias, finite data and detectors, computational bounds, and incomplete verification itself—and describe the physicist’s future role as designer, monitor, and evaluator who encodes scientific rigor into increasingly autonomous systems. The practical payoff is a context-dependent verification checklist that lets physicists deploy ML aggressively without corrupting statistical claims.

Core claim

Verification of machine learning is essential precisely when its outputs form part of the statistical model used for inference or hypothesis testing; at other stages of the discovery workflow imperfect models are acceptable provided residual uncertainties are quantified and systematic biases are either calibrated out or demonstrably absent.

What carries the argument

The four-stage statistical workflow (data collection, summarization, modeling, inference) together with the aleatoric/epistemic uncertainty distinction; these locate every ML tool and dictate whether, and which, verification is required.

Load-bearing premise

That the four-stage workflow and the aleatoric/epistemic split cleanly cover every present and future machine-learning use in fundamental physics, so the verification rules derived from them stay complete.

What would settle it

A concrete ML application whose outputs enter a discovery claim yet cannot be classified as either (a) a summarization/exploratory step whose imperfections only reduce power or (b) a modeling/surrogate step whose residual bias and epistemic uncertainty can be quantified and propagated, thereby leaving the paper’s decision rules incomplete.

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

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

0 major / 5 minor

Summary. This VERaiPHY community review argues that ML verification in fundamental physics is essential precisely when model outputs enter statistical modeling, inference, or hypothesis testing, while imperfect models remain tolerable in summarization, exploratory analysis, and calibrated surrogates provided residual uncertainties are quantified and unmodeled systematic bias is avoided. It situates ML within a four-stage discovery workflow (data collection, summarization, modeling, inference), surveys computational bottlenecks and emerging paradigms (differentiable design, foundation models, anomaly detection, agentic AI), and articulates irreducible limits (inductive bias, observational constraints, computational bounds, verification incompleteness). The closing sections discuss the physicist’s evolving role as designer, evaluator, and teacher of AI systems and offer high-level guidelines for responsible deployment.

Significance. As a synthesis paper rather than a primary-result claim, its value lies in organizing a fragmented literature into a coherent, workflow-based verification framework that spans particle physics, astrophysics, and cosmology. The contextual distinction between performance degradation and statistical invalidity (Sections 3.1–3.3), the explicit treatment of agentic systems and verification limits (2.3, 4.4), and the reflection on human oversight (Section 5) are timely contributions for a community facing increasingly autonomous ML. The paper correctly grounds its arguments in standard statistical practice (look-elsewhere effects, calibration, coverage) and citable results (No Free Lunch, data-processing inequality, SBI surveys). It does not overclaim completeness and is well positioned as an entry point to the broader VERaiPHY series.

minor comments (5)
  1. Several companion VERaiPHY reviews are cited as “in preparation” (e.g., Refs. [36], [61], [103], [114], [118]). For archival permanence, either update with arXiv identifiers where available or flag more clearly which claims rest only on forthcoming companion pieces.
  2. Figure 1 and Figure 3 are conceptually clear but would benefit from slightly more explicit captions linking each panel to the corresponding workflow stage or uncertainty type discussed in the text.
  3. Section 2.3 on agentic AI is appropriately cautious; a short forward pointer to concrete verification protocols (even if only as open problems) would strengthen the bridge to Section 4.4.
  4. Minor typographical and formatting inconsistencies appear (e.g., spacing around citations, occasional hyphenation of “black-box” / “black boxes”). A light copy-edit pass would polish the manuscript.
  5. The abstract and concluding guidelines are strong; ensuring the five bullet guidelines in Section 6 map one-to-one onto the section structure would improve navigability for practitioners.

Circularity Check

0 steps flagged

No circularity: community review with normative framing, no fitted predictions or self-definitional reductions.

full rationale

This is a VERaiPHY community review that organizes existing statistical practice around a four-stage workflow (data collection, summarization, modeling, inference) and an aleatoric/epistemic taxonomy. It does not claim to derive quantitative predictions, uniqueness theorems, or first-principles results from fitted parameters or self-defined quantities. Load-bearing external anchors (Wolpert No Free Lunch, Cover data-processing inequality, Cranmer SBI survey, standard frequentist/Bayesian practice) are independent of the authors. Companion VERaiPHY citations supply depth on subtopics but are not required to force the central normative claim that verification is essential precisely when ML outputs enter statistical modeling, inference, or hypothesis testing. No equation reduces to its own input by construction; no ansatz is smuggled via self-citation; no known empirical pattern is merely renamed. Score 0 is the correct honest finding.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

As a review the paper introduces no free parameters and no new physical entities. Its load-bearing premises are standard mathematical results and domain conventions of experimental particle physics, astrophysics, and cosmology that are explicitly invoked to justify the verification rules.

axioms (4)
  • standard math No Free Lunch theorems: no learning algorithm is universally superior; inductive bias is unavoidable.
    Invoked in Section 4.1 to argue that physicists must choose and justify bias rather than hope for bias-free models.
  • standard math Data Processing Inequality: deterministic or stochastic transformations cannot increase mutual information with the quantity of interest.
    Used in Section 4.2 to bound what data augmentation can achieve.
  • domain assumption Discovery in fundamental physics proceeds via statistical inference on noisy, incomplete observables rather than direct observation of the target entities.
    Stated in the Introduction and Section 2; underpins the entire claim that ML must preserve statistical validity.
  • domain assumption The four-stage workflow (collection, summarization, modeling, inference) plus the aleatoric/epistemic distinction covers the relevant ML insertion points.
    Introduced via Figure 1 and Section 2.1; all subsequent 'when verification matters' rules are derived from this partition.

pith-pipeline@v1.1.0-grok45 · 29995 in / 2220 out tokens · 21184 ms · 2026-07-14T00:49:47.655260+00:00 · methodology

0 comments
read the original abstract

Machine learning (ML) has become integral to fundamental physics, accelerating statistical workflows from data acquisition through inference and hypothesis testing. As ML systems grow increasingly autonomous, ensuring their reliability for discovery claims becomes critical. This review synthesizes the VERaiPHY (Validation & Evaluation for Robust AI in PHYsics) initiative's frameworks for rigorous ML assessment across particle physics, astrophysics, and cosmology. We establish when verification is essential by contextualizing ML within the statistical discovery workflow. We emphasize fundamental limitations: inductive bias is unavoidable, sample complexity bounds learning, and experimental constraints limit discovery. We reflect on physicists' evolving role as both experimental designers and evaluators whose judgments encode scientific rigor into AI systems. Responsible integration requires understanding ML's transformative potential alongside its intrinsic boundaries.

Figures

Figures reproduced from arXiv: 2607.10039 by Gaia Grosso, Lukas Heinrich, Vinicius Mikuni.

Figure 1
Figure 1. Figure 1: The statistical workflow of fundamental physics experiments. The natural phenomenon is observed using detectors. The collected raw data are trans￾formed into summary statistics (s(x)), that are then modeled and used for statistical inference tasks including parameter estimation and statistical tests. uring interferometer sensitivities and data acquisition systems. This stage is largely ir￾reversible: infor… view at source ↗
Figure 2
Figure 2. Figure 2: Two different physical phe￾nomena (top row and bottom row) give rise to two different signatures in the observable data representations (on the right side). Simulations al￾low to probe these properties, “fold￾ing" theories into the observable data representations, to design and vali￾date the statistical workflow before analyzing real data. This is precisely why simulation is not merely con￾venient but nece… view at source ↗
Figure 3
Figure 3. Figure 3: When do we need to estimate uncertainties and which kind. (a) Aleatoric uncertainties must be propagated from the collected data all the way to the discovery claim; (b) Epistemic uncertainties affecting the statistical model of the summary statistics must be included in the model as well; (c) the epistemic uncer￾tainties affecting a surrogate model replacing a physical model for data generation must be pro… view at source ↗
Figure 4
Figure 4. Figure 4: The physicist of the Future. We lean toward orchestration, monitor and verification of increasingly automatic pipelines. inevitably shift. Rather than spending most of their effort on implementing and optimizing statistical procedures, physicists and statisticians will devote more attention to formulating the right scientific questions, defining meaningful hypotheses, and determining which results should b… view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

142 extracted references · 6 canonical work pages · 3 internal anchors

  1. [1]

    2025 , month =

    Cho, Kyunghyun , title =. 2025 , month =

    2025

  2. [2]

    Coffea: Columnar Object Framework For Effective Analysis

    Smith, Nicholas and others. Coffea: Columnar Object Framework For Effective Analysis. EPJ Web Conf. 2020. doi:10.1051/epjconf/202024506012. arXiv:2008.12712

    doi:10.1051/epjconf/202024506012 2020

  3. [3]

    Proceedings of the 24th International Conference on Computing in High Energy and Nuclear Physics (CHEP 2019) , journal =

    Awkward Arrays in Python, C++, and Numba , author =. Proceedings of the 24th International Conference on Computing in High Energy and Nuclear Physics (CHEP 2019) , journal =. 2020 , doi =

    2019

  4. [4]

    and Rademakers, F

    Brun, R. and Rademakers, F. ROOT: An object oriented data analysis framework. Nucl. Instrum. Meth. A. 1997. doi:10.1016/S0168-9002(97)00048-X

    doi:10.1016/s0168-9002(97)00048-x 1997

  5. [5]

    Active Learning reinterpretation of an ATLAS Dark Matter search constraining a model of a dark Higgs boson decaying to two b-quarks. 2022

    2022

  6. [6]

    Observation of Gravitational Waves from a Binary Black Hole Merger , author =. Phys. Rev. Lett. , volume =. 2016 , month =. doi:10.1103/PhysRevLett.116.061102 , url =

    doi:10.1103/physrevlett.116.061102 2016

  7. [7]

    The Astronomical Journal , volume=

    Data release 1 of the dark energy spectroscopic instrument , author=. The Astronomical Journal , volume=. 2026 , publisher=

    2026

  8. [8]

    Monthly Notices of the Royal Astronomical Society , volume=

    Cosmic confusion: degeneracies among cosmological parameters derived from measurements of microwave background anisotropies , author=. Monthly Notices of the Royal Astronomical Society , volume=. 1999 , publisher=

    1999

  9. [9]

    Monthly Notices of the Royal Astronomical Society , volume=

    Massively parallel Bayesian inference for transient gravitational-wave astronomy , author=. Monthly Notices of the Royal Astronomical Society , volume=. 2020 , publisher=

    2020

  10. [10]

    Toward Early-Warning Detection of Gravitational Waves from Compact Binary Coalescence

    Cannon, Kipp and others. Toward Early-Warning Detection of Gravitational Waves from Compact Binary Coalescence. Astrophys. J. 2012. doi:10.1088/0004-637X/748/2/136. arXiv:1107.2665

    Pith/arXiv arXiv doi:10.1088/0004-637x/748/2/136 2012

  11. [11]

    Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC

    Aad, Georges and others. Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC. Phys. Lett. B. 2012. doi:10.1016/j.physletb.2012.08.020. arXiv:1207.7214

    Pith/arXiv arXiv doi:10.1016/j.physletb.2012.08.020 2012

  12. [12]

    Observation of a New Boson at a Mass of 125 GeV with the CMS Experiment at the LHC

    Chatrchyan, Serguei and others. Observation of a New Boson at a Mass of 125 GeV with the CMS Experiment at the LHC. Phys. Lett. B. 2012. doi:10.1016/j.physletb.2012.08.021. arXiv:1207.7235

    Pith/arXiv arXiv doi:10.1016/j.physletb.2012.08.021 2012

  13. [13]

    Asymptotic formulae for likelihood-based tests of new physics

    Cowan, Glen and Cranmer, Kyle and Gross, Eilam and Vitells, Ofer. Asymptotic formulae for likelihood-based tests of new physics. Eur. Phys. J. C. 2011. doi:10.1140/epjc/s10052-011-1554-0. arXiv:1007.1727

    Pith/arXiv arXiv doi:10.1140/epjc/s10052-011-1554-0 2011

  14. [14]

    Journal of Instrumentation , volume=

    The CMS trigger system , author=. Journal of Instrumentation , volume=

  15. [15]

    Journal of Instrumentation , volume=

    Operation of the ATLAS trigger system in Run 2 , author=. Journal of Instrumentation , volume=

  16. [16]

    arXiv preprint arXiv:1902.08960 , year=

    Precision laser-based measurements of the single electron response of SPCs for the NEWS-G light dark matter search experiment , author=. arXiv preprint arXiv:1902.08960 , year=

    Pith/arXiv arXiv 1902

  17. [17]

    Physical Review D , volume=

    First results from the CRESST-III low-mass dark matter program , author=. Physical Review D , volume=. 2019 , publisher=

    2019

  18. [18]

    Fast neural-net based fake track rejection in the LHCb reconstruction , author=

  19. [19]

    The Phase-2 Upgrade of the CMS Level-1 Trigger

    Zabi, Alexandre and Berryhill, Jeffrey Wayne and Perez, Emmanuelle and Tapper, Alexander D. The Phase-2 Upgrade of the CMS Level-1 Trigger. 2020

    2020

  20. [20]

    Journal of instrumentation , volume=

    Fast inference of deep neural networks in FPGAs for particle physics , author=. Journal of instrumentation , volume=. 2018 , publisher=

    2018

  21. [21]

    arXiv preprint arXiv:2509.24371 , year=

    Advancing the CMS Level-1 Trigger: Jet Tagging with DeepSets at the HL-LHC , author=. arXiv preprint arXiv:2509.24371 , year=

    arXiv

  22. [22]

    Autoencoders on field-programmable gate arrays for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider

    Govorkova, Ekaterina and others. Autoencoders on field-programmable gate arrays for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider. Nature Mach. Intell. 2022. doi:10.1038/s42256-022-00441-3. arXiv:2108.03986

    Pith/arXiv arXiv doi:10.1038/s42256-022-00441-3 2022

  23. [23]

    Matching Matched Filtering with Deep Networks for Gravitational-Wave Astronomy , author =. Phys. Rev. Lett. , volume =. 2018 , month =. doi:10.1103/PhysRevLett.120.141103 , url =

    work page doi:10.1103/physrevlett.120.141103 2018

  24. [24]

    Physical Review D , volume=

    Generalized approach to matched filtering using neural networks , author=. Physical Review D , volume=. 2022 , publisher=

    2022

  25. [25]

    Physical Review D , volume=

    Deep neural networks to enable real-time multimessenger astrophysics , author=. Physical Review D , volume=. 2018 , publisher=

    2018

  26. [26]

    Machine Learning: Science and Technology , volume=

    GWAK: gravitational-wave anomalous knowledge with recurrent autoencoders , author=. Machine Learning: Science and Technology , volume=. 2024 , publisher=

    2024

  27. [27]

    SEAL - A Symmetry EncourAging Loss for High Energy Physics

    Hebbar, Pradyun and Madula, Thandikire and Mikuni, Vinicius and Nachman, Benjamin and Outmezguine, Nadav and Savoray, Inbar. SEAL - A Symmetry EncourAging Loss for High Energy Physics. 2025. arXiv:2511.01982

    arXiv 2025

  28. [28]

    Machine Learning: Science and Technology , volume=

    Probing the effects of broken symmetries in machine learning , author=. Machine Learning: Science and Technology , volume=. 2024 , publisher=

    2024

  29. [29]

    Physical Review D , volume=

    Learning broken symmetries with approximate invariance , author=. Physical Review D , volume=. 2025 , publisher=

    2025

  30. [30]

    and Ipp, Andreas and M

    Aarts, Gert and Habibi, Diaa E. and Ipp, Andreas and M. Generalizable Equivariant Diffusion Models for Non-Abelian Lattice Gauge Theory. 2026. arXiv:2601.19552

    arXiv 2026

  31. [31]

    A Living Review of Machine Learning for Particle Physics

    Feickert, Matthew and Nachman, Benjamin. A Living Review of Machine Learning for Particle Physics. 2021. arXiv:2102.02770

    Pith/arXiv arXiv 2021

  32. [32]

    OmniJet- : The first cross-task foundation model for particle physics

    Birk, Joschka and Hallin, Anna and Kasieczka, Gregor. OmniJet- : The first cross-task foundation model for particle physics. 2024. arXiv:2403.05618

    Pith/arXiv arXiv 2024

  33. [33]

    Foundation models for high-energy physics

    Hallin, Anna. Foundation models for high-energy physics. 2nd European AI for Fundamental Physics Conference. 2025. arXiv:2509.21434

    arXiv 2025

  34. [34]

    Re-Simulation-based Self-Supervised Learning for Pre-Training Foundation Models

    Harris, Philip and Kagan, Michael and Krupa, Jeffrey and Maier, Benedikt and Woodward, Nathaniel. Re-Simulation-based Self-Supervised Learning for Pre-Training Foundation Models. 2024. arXiv:2403.07066

    Pith/arXiv arXiv 2024

  35. [35]

    Masked particle modeling on sets: towards self-supervised high energy physics foundation models

    Golling, Tobias and Heinrich, Lukas and Kagan, Michael and Klein, Samuel and Leigh, Matthew and Osadchy, Margarita and Raine, John Andrew. Masked particle modeling on sets: towards self-supervised high energy physics foundation models. Mach. Learn. Sci. Tech. 2024. doi:10.1088/2632-2153/ad64a8. arXiv:2401.13537

    Pith/arXiv arXiv doi:10.1088/2632-2153/ad64a8 2024

  36. [36]

    Finetuning foundation models for joint analysis optimization in High Energy Physics

    Vigl, Matthias and Hartman, Nicole and Heinrich, Lukas. Finetuning foundation models for joint analysis optimization in High Energy Physics. Mach. Learn. Sci. Tech. 2024. doi:10.1088/2632-2153/ad55a3. arXiv:2401.13536

    Pith/arXiv arXiv doi:10.1088/2632-2153/ad55a3 2024

  37. [37]

    Method to simultaneously facilitate all jet physics tasks

    Mikuni, Vinicius and Nachman, Benjamin. Method to simultaneously facilitate all jet physics tasks. Phys. Rev. D. 2025. doi:10.1103/PhysRevD.111.054015. arXiv:2502.14652

    doi:10.1103/physrevd.111.054015 2025

  38. [38]

    Solving key challenges in collider physics with foundation models

    Mikuni, Vinicius and Nachman, Benjamin. Solving key challenges in collider physics with foundation models. Phys. Rev. D. 2025. doi:10.1103/PhysRevD.111.L051504. arXiv:2404.16091

    doi:10.1103/physrevd.111.l051504 2025

  39. [39]

    Is Tokenization Needed for Masked Particle Modelling?

    Leigh, Matthew and Klein, Samuel and Charton, Fran c ois and Golling, Tobias and Heinrich, Lukas and Kagan, Michael and Ochoa, In \^e s and Osadchy, Margarita. Is Tokenization Needed for Masked Particle Modelling?. 2024. arXiv:2409.12589

    Pith/arXiv arXiv 2024

  40. [40]

    HEP-JEPA: A foundation model for collider physics using joint embedding predictive architecture

    Bardhan, Jai and Agrawal, Radhikesh and Tilak, Abhiram and Neeraj, Cyrin and Mitra, Subhadip. HEP-JEPA: A foundation model for collider physics using joint embedding predictive architecture. 2025. arXiv:2502.03933

    Pith/arXiv arXiv 2025

  41. [41]

    Particle transformers for identifying Lorentz-boosted Higgs bosons decaying to a pair of W bosons

    Hayrapetyan, Aram and others. Particle transformers for identifying Lorentz-boosted Higgs bosons decaying to a pair of W bosons. 2026. arXiv:2604.09809

    Pith/arXiv arXiv 2026

  42. [42]

    OmniCosmos: Transferring Particle Physics Knowledge Across the Cosmos

    Mikuni, Vinicius and Elsharkawy, Ibrahim and Nachman, Benjamin. OmniCosmos: Transferring Particle Physics Knowledge Across the Cosmos. 2025. arXiv:2512.24422

    arXiv 2025

  43. [43]

    OmniMol: Transferring Particle Physics Knowledge to Molecular Dynamics with Point-Edge Transformers

    Elsharkawy, Ibrahim and Mikuni, Vinicius and Bhimji, Wahid and Nachman, Benjamin. OmniMol: Transferring Particle Physics Knowledge to Molecular Dynamics with Point-Edge Transformers. 2026. arXiv:2601.10791

    Pith/arXiv arXiv 2026

  44. [44]

    PASCL: Supervised Contrastive Learning with Perturbative Augmentation for Particle Decay Reconstruction

    Lu, Junjian and Liu, Siwei and Kobylianskii, Dmitrii and Dreyer, Etienne and Gross, Eilam and Liang, Shangsong. PASCL: supervised contrastive learning with perturbative augmentation for particle decay reconstruction. Mach. Learn. Sci. Tech. 2024. doi:10.1088/2632-2153/ad8060. arXiv:2402.11538

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/2632-2153/ad8060 2024

  45. [46]

    Advances in Neural Information Processing Systems , volume=

    Autoscidact: Automated scientific discovery through contrastive embedding and hypothesis testing , author=. Advances in Neural Information Processing Systems , volume=

  46. [47]

    MACK: Mismodeling Addressed with Contrastive Knowledge

    Sheldon, Liam Rankin and Rankin, Dylan Sheldon and Harris, Philip. MACK: Mismodeling addressed with contrastive knowledge. SciPost Phys. 2025. doi:10.21468/SciPostPhys.18.5.150. arXiv:2410.13947

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.21468/scipostphys.18.5.150 2025

  47. [48]

    A cautionary tale of decorrelating theory uncertainties

    Ghosh, Aishik and Nachman, Benjamin. A cautionary tale of decorrelating theory uncertainties. Eur. Phys. J. C. 2022. doi:10.1140/epjc/s10052-022-10012-w. arXiv:2109.08159

    Pith/arXiv arXiv doi:10.1140/epjc/s10052-022-10012-w 2022

  48. [49]

    A unified approach for jet tagging in Run 3 at s =13.6 TeV in CMS. 2024

    2024

  49. [50]

    Transforming jet flavour tagging at ATLAS

    Aad, Georges and others. Transforming jet flavour tagging at ATLAS. Nature Commun. 2026. doi:10.1038/s41467-025-65059-6. arXiv:2505.19689

    doi:10.1038/s41467-025-65059-6 2026

  50. [51]

    SciPost Physics , volume=

    Machine learning and LHC event generation , author=. SciPost Physics , volume=

  51. [52]

    CaloChallenge 2022: a community challenge for fast calorimeter simulation

    Amram, Oz and others. CaloChallenge 2022: a community challenge for fast calorimeter simulation. Rept. Prog. Phys. 2025. doi:10.1088/1361-6633/ae1304. arXiv:2410.21611

    doi:10.1088/1361-6633/ae1304 2022

  52. [53]

    arXiv preprint arXiv:2110.07925 , year=

    Machine learning for the LHCb simulation , author=. arXiv preprint arXiv:2110.07925 , year=

    Pith/arXiv arXiv

  53. [54]

    Machine Learning: Science and Technology , volume=

    Particle-based fast jet simulation at the LHC with variational autoencoders , author=. Machine Learning: Science and Technology , volume=. 2022 , publisher=

    2022

  54. [55]

    2024 , institution=

    The Fast Simulation Program of ATLAS at the LHC , author=. 2024 , institution=

    2024

  55. [56]

    arXiv preprint arXiv:2511.02020 , year=

    Machine learning in LHCb Simulation: From fast to flash , author=. arXiv preprint arXiv:2511.02020 , year=

    arXiv

  56. [57]

    Proceedings of the National Academy of Sciences , volume =

    Siyu He and Yin Li and Yu Feng and Shirley Ho and Siamak Ravanbakhsh and Wei Chen and Barnabás Póczos , title =. Proceedings of the National Academy of Sciences , volume =. 2019 , doi =

    2019

  57. [58]

    Computational Astrophysics and Cosmology , volume=

    Fast cosmic web simulations with generative adversarial networks , author=. Computational Astrophysics and Cosmology , volume=. 2018 , publisher=

    2018

  58. [59]

    Machine Learning: Science and Technology , url=

    Cappelli, Pietro and Grosso, Gaia and Letizia, Marco and Reyes-González, Humberto and Zanetti, Marco , title=. Machine Learning: Science and Technology , url=

  59. [60]

    Evaluating generative models in high energy physics

    Kansal, Raghav and Li, Anni and Duarte, Javier and Chernyavskaya, Nadezda and Pierini, Maurizio and Orzari, Breno and Tomei, Thiago. Evaluating generative models in high energy physics. Phys. Rev. D. 2023. doi:10.1103/PhysRevD.107.076017. arXiv:2211.10295

    Pith/arXiv arXiv doi:10.1103/physrevd.107.076017 2023

  60. [61]

    hydrodynamical simulations , author=

    Machine learning and cosmological simulations--ii. hydrodynamical simulations , author=. Monthly Notices of the Royal Astronomical Society , volume=. 2016 , publisher=

    2016

  61. [62]

    Proceedings of the National Academy of Sciences , volume=

    The frontier of simulation-based inference , author=. Proceedings of the National Academy of Sciences , volume=. 2020 , publisher=

    2020

  62. [63]

    Artificial Intelligence for High Energy Physics , pages=

    Simulation-based inference methods for particle physics , author=. Artificial Intelligence for High Energy Physics , pages=. 2022 , publisher=

    2022

  63. [64]

    Reports on Progress in Physics , volume=

    An implementation of neural simulation-based inference for parameter estimation in ATLAS , author=. Reports on Progress in Physics , volume=. 2025 , publisher=

    2025

  64. [65]

    Machine Learning: Science and Technology , volume=

    Robust simulation-based inference in cosmology with Bayesian neural networks , author=. Machine Learning: Science and Technology , volume=. 2023 , publisher=

    2023

  65. [66]

    Nature Astronomy , volume=

    A deep-learning algorithm to disentangle self-interacting dark matter and AGN feedback models , author=. Nature Astronomy , volume=. 2024 , publisher=

    2024

  66. [67]

    Machine Learning: Science and Technology , abstract =

    Astrand, S and Boggia, L and Borsato, M and Bozianu, L and Cocha Toapaxi, C E and Giasemis, F I and Hansen, J and Inkaew, P and Iversen, K E and Jawahar, P and Pineiro Monteagudo, H and Olocco, M and Schramm, S , title =. Machine Learning: Science and Technology , abstract =. 2026 , month =. doi:10.1088/2632-2153/ae35cc , url =

    work page doi:10.1088/2632-2153/ae35cc 2026

  67. [68]

    arXiv preprint arXiv:1702.08608 , year=

    Towards a rigorous science of interpretable machine learning , author=. arXiv preprint arXiv:1702.08608 , year=

    Pith/arXiv arXiv

  68. [69]

    Nature machine intelligence , volume=

    Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead , author=. Nature machine intelligence , volume=. 2019 , publisher=

    2019

  69. [70]

    , author=

    The mythos of model interpretability: In machine learning, the concept of interpretability is both important and slippery. , author=. Queue , volume=. 2018 , publisher=

    2018

  70. [71]

    arXiv preprint arXiv:2601.03220 , year=

    From Entropy to Epiplexity: Rethinking Information for Computationally Bounded Intelligence , author=. arXiv preprint arXiv:2601.03220 , year=

    arXiv

  71. [72]

    arXiv preprint arXiv:2503.02113 , year=

    Deep learning is not so mysterious or different , author=. arXiv preprint arXiv:2503.02113 , year=

    Pith/arXiv arXiv

  72. [73]

    arXiv preprint arXiv:2304.05366 , year=

    The no free lunch theorem, kolmogorov complexity, and the role of inductive biases in machine learning , author=. arXiv preprint arXiv:2304.05366 , year=

    Pith/arXiv arXiv

  73. [74]

    Neural computation , volume=

    The lack of a priori distinctions between learning algorithms , author=. Neural computation , volume=. 1996 , publisher=

    1996

  74. [75]

    1995 , institution=

    No free lunch theorems for search , author=. 1995 , institution=

    1995

  75. [76]

    IEEE transactions on evolutionary computation , volume=

    No free lunch theorems for optimization , author=. IEEE transactions on evolutionary computation , volume=. 2002 , publisher=

    2002

  76. [77]

    arXiv preprint arXiv:1805.08522 , year=

    Deep learning generalizes because the parameter-function map is biased towards simple functions , author=. arXiv preprint arXiv:1805.08522 , year=

    Pith/arXiv arXiv

  77. [78]

    arXiv preprint arXiv:2103.10427 , year=

    The low-rank simplicity bias in deep networks , author=. arXiv preprint arXiv:2103.10427 , year=

    Pith/arXiv arXiv

  78. [79]

    arXiv preprint arXiv:2509.22445 , year=

    Bridging Kolmogorov Complexity and Deep Learning: Asymptotically Optimal Description Length Objectives for Transformers , author=. arXiv preprint arXiv:2509.22445 , year=

    arXiv

  79. [80]

    Resimulation-based self-supervised learning for pretraining physics foundation models , author =. Phys. Rev. D , volume =. 2025 , month =. doi:10.1103/PhysRevD.111.032010 , url =

    doi:10.1103/physrevd.111.032010 2025

  80. [81]

    Anomaly-preserving contrastive neural embeddings for end-to-end model-independent searches at the LHC , author =. Phys. Rev. D , volume =. 2025 , month =. doi:10.1103/5n77-ynsp , url =

    work page doi:10.1103/5n77-ynsp 2025

Showing first 80 references.