Pith. sign in

REVIEW 2 major objections 6 minor 55 references

Reviewed by Pith at T0; open to challenge.

T0 means a machine referee read the full paper against a public rubric. The mark states how deep the mechanical check went, never who wrote it. the ladder, T0–T4 →

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

T0 review · glm-5.2

Train classifiers on unlabeled mixtures to find exotic gamma-ray signals

2026-07-09 18:37 UTC pith:L4WZ4ZAL

load-bearing objection Honest proof-of-concept for BvM weak supervision in gamma-ray astronomy; math is sound, case studies are well-designed, but the same-background assumption is untested under realistic conditions the 2 major comments →

arxiv 2607.07158 v1 pith:L4WZ4ZAL submitted 2026-07-08 astro-ph.HE astro-ph.IMhep-ph

Weakly supervised machine learning for model-agnostic searches of new phenomena in the γ-ray sky

Michael Kr\"amer , Silvia Manconi , Kathrin Nippel This is my paper
classification astro-ph.HE astro-ph.IMhep-ph PACS 95.75.Pq95.85.Pw95.30.Cq
keywords signalsupervisedgammaclassificationfullyphenomenasearchesweakly
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 proposes that classifiers trained to distinguish a pure background sample from a mixed sample containing an unknown fraction of signal can recover signal-like events without ever being shown labeled signal examples. The mathematical core is simple: if the ratio of mixed-sample to background-sample densities is f times the signal-to-background likelihood ratio plus (1-f), then for any nonzero signal fraction f, this ratio is monotonically related to the true signal likelihood ratio. A classifier trained on the two samples therefore learns a score that ranks events by how signal-like they are, even though no individual event was ever labeled as signal. The authors test this background-versus-mixture approach on three gamma-ray astrophysics problems: separating pulsars from active galactic nuclei (a clean benchmark where the method approaches fully supervised performance), identifying dark matter subhalos (where performance is competitive but degraded by spectral overlap with astrophysical sources), and detecting spectral wiggles induced by axion-photon oscillations (where both supervised and weakly supervised methods struggle when the modulation amplitude is small). In each case, the signal fraction in the mixed sample controls the tradeoff between true positive rate and false positive rate: larger fractions improve sensitivity but admit more background contamination. The method is positioned not as a replacement for model-specific likelihood searches but as a model-agnostic candidate-selection and anomaly-ranking tool that can flag interesting sources for follow-up without committing to a particular signal hypothesis during training.

Core claim

A classifier trained to separate a pure background sample from a mixed sample containing an unknown signal fraction learns a decision score that is monotonically related to the optimal signal-versus-background likelihood ratio, because the density ratio p_M(x)/p_B(x) = f * p_E(x)/p_A(x) + (1-f) is strictly increasing in p_E(x)/p_A(x) for any f in (0,1]. This means weakly supervised training on unlabeled mixtures can rank gamma-ray sources by how exotic they are, approaching fully supervised performance when signal and background are well separated, without requiring labeled signal examples during training.

What carries the argument

The background-versus-mixture (BvM) setup: two training samples are constructed, one pure background B drawn from p_A(x) and one mixture M drawn from f*p_E(x) + (1-f)*p_A(x). A boosted decision tree is trained to distinguish B from M. Because the background component is identically distributed in both samples, the classifier implicitly learns the signal-to-background density ratio p_E(x)/p_A(x). The signal fraction f and the relative sample sizes control the conservatism of the decision boundary. Normalizing flows are used to augment background samples when the catalog contains too few real sources for controlled mixture studies.

Load-bearing premise

The astrophysical background component must be distributed identically in the pure background sample and the mixed sample. All three case studies satisfy this by construction using simulated or carefully matched data, but in a real application to unassociated Fermi-LAT sources, selection effects such as differing exposure, Galactic latitude, or detection-significance thresholds between identified and unidentified sources would likely violate this assumption and bias the clasS

What would settle it

If the background distributions in the pure background sample and the mixed sample differ for reasons other than signal admixture (e.g., selection effects, exposure differences, or correlations between mixture-defining variables and input features), the classifier score is no longer monotone in the signal-to-background likelihood ratio, and the method produces biased rankings. The paper acknowledges this but validates only on controlled samples where the same-background assumption holds by construction.

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

If this is right

  • Weakly supervised classifiers could be applied directly to unassociated Fermi-LAT sources, using identified astrophysical sources as background and unassociated sources as the mixture, to produce a ranked candidate list for follow-up observations without assuming a specific dark matter or new physics model.
  • The same-background assumption means that any selection effects differing between identified and unassociated source populations (exposure, Galactic latitude, flux thresholds) would bias the classifier, so real-data application requires careful matching or reweighting of the background sample to the mixture sample.
  • The method generalizes to any spectral anomaly search where a reference population of normal sources can be defined, including TeV gamma-ray spectra from Cherenkov telescopes or X-ray observations, provided the energy binning is fine enough to capture the relevant spectral features.
  • The tradeoff between signal fraction and false positive rate implies that optimal candidate selection may require scanning over multiple mixture constructions or combining weakly supervised scores from several signal-fraction settings.

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

2 major / 6 minor

Summary. The manuscript explores weakly supervised classification in a background-versus-mixture (BvM) setup for model-agnostic searches of new phenomena in Fermi-LAT gamma-ray data. The central mathematical claim—that the BvM classifier score is monotone in the signal-to-background likelihood ratio p_E(x)/p_A(x) when the background distribution is identical across samples—is a standard CWoLa result and is correctly stated. Three case studies of increasing difficulty are presented: pulsar-AGN separation (controlled benchmark), dark-matter subhalo identification, and ALP-induced spectral irregularities. In each case, weakly supervised performance is compared against a supervised baseline, and the dependence on signal fraction and sample composition is characterized. The approach is positioned not as a replacement for likelihood-based discovery, but as a candidate-selection and anomaly-ranking strategy. The presentation is clear and the physics motivations are well developed.

Significance. The application of weakly supervised (CWoLa-type) methods to gamma-ray source classification is timely and has not, to the authors' knowledge, been previously attempted in this context. The mathematical foundation (Eq. 2.1 and the monotonicity argument in §2.2) is sound and clearly stated. The three case studies are well-chosen, spanning a controlled benchmark, a model-driven exotic search, and a subtle spectral-deformation scenario, which collectively illustrate both the potential and the limitations of the method. The honest reporting of performance degradation relative to supervised baselines—particularly in the ALP case where the signal is subtle—adds credibility. The normalizing-flow augmentation of the AGN background (Appendix A) is a useful technical contribution, validated with a classifier-discrimination test. The work is a reasonable proof-of-concept that connects developments in collider-physics anomaly detection to high-energy astrophysics.

major comments (2)
  1. §2.2 and §6: The same-background assumption—that p_A(x) is identically distributed in the background sample B and the mixed sample M—is the load-bearing condition for the monotonicity result. All three case studies satisfy this by construction: the pulsar-AGN benchmark uses flow-augmented AGN from the same learned distribution; the DM subhalo case injects signal into a disjoint subset of the same background pool; the ALP case applies modulations to randomly chosen spectra from the same simulated pulsar population. The authors acknowledge that realistic applications to unassociated Fermi-LAT sources would face selection-effect violations (§2.2, §4.4, §6) and defer this to future work. This is an honest framing, but it means the paper demonstrates the method under idealized conditions only. The central claim—that weak supervision 'can identify anomalous or signal-like subsets of data' (§6,
  2. §3.4: The background sample B is augmented with normalizing-flow-generated AGN spectra, while the mixed sample M contains real AGN (plus pulsars). The same-background assumption requires that the flow-generated and real AGN spectra follow the same distribution. Appendix A validates the flow with a classifier-discrimination test (accuracy 52.6%, AUC 0.53), which is reassuring. However, the flow is trained on the same 4FGL AGN population used to construct M, so any subtle distributional mismatch between generated and real spectra would directly bias the BvM classifier in a way that is hard to detect from the discrimination test alone. The authors should discuss whether the flow-generated AGN in B and the real AGN in M are drawn from statistically independent realizations, or whether there is overlap, and whether the discrimination test is sensitive enough to detect the level of mismatch
minor comments (6)
  1. §3.3: The statement 'Adding positional or flux-history information does not further improve performance for BDTs' could benefit from a quantitative comparison (e.g., TPR/FPR values) to support the claim, since the preceding paragraph gives specific numbers for the flux-band-only and all-features cases.
  2. Table 1: For g_aγ = 50×10^{-11} GeV^{-1}, the TPR values (0.157 and 0.108) are very low. The text in §5.3 notes this is expected, but it would help to state explicitly in the table caption that these values indicate the classifier is essentially failing to identify modulated spectra at this coupling, rather than underperforming.
  3. §5.2: The choice m_a = 1 neV is mentioned without justification in the main text; the reader must consult Appendix B. A one-sentence motivation in §5.2 would improve readability.
  4. Figure 3: The arrow convention (tail = supervised, tip = weakly supervised) is explained in the caption but is somewhat unusual. Consider adding a small legend or making the convention more visually intuitive.
  5. §4.4, last paragraph: The sentence 'Improved separation may be expected for annihilation channels with harder spectra, for example τ+τ−' is reasonable but reads as speculation without supporting evidence. A brief quantitative comparison or a reference would strengthen it.
  6. The abstract states 'in favourable cases, the method approaches the performance of fully supervised classifiers.' Given that this is true primarily for the pulsar-AGN benchmark and not for the more physically motivated DM or ALP cases, a slightly more qualified phrasing would be more precise.

Circularity Check

0 steps flagged

No circularity found: the central mathematical claim is a standard probability result, and self-citations serve only as reference benchmarks.

full rationale

The paper's load-bearing mathematical claim (§2.2) is that p_M(x)/p_B(x) = f·p_E(x)/p_A(x) + (1−f) is strictly increasing in the likelihood ratio p_E(x)/p_A(x) for f ∈ (0,1]. This is a standard, self-contained probability identity that does not depend on the authors' prior work. The supervised benchmarks in §3.3 and §4.3 are compared against the authors' own prior results ([14], [20]), but these are used as performance reference points, not as inputs to the weak-supervision derivation. The normalizing flow (Appendix A) is used only for data augmentation of background samples and is validated independently (a classifier distinguishing real from generated spectra achieves 52.6% accuracy, consistent with random guessing). The three case studies (pulsar-AGN, dark matter subhalos, ALP modulations) each construct controlled test scenarios where the same-background assumption is satisfied by construction, and the authors explicitly acknowledge this as a limitation for real-world application (§2.2, §4.4, §6). No step in the derivation chain reduces to its own inputs by definition, no prediction is a renamed fit, and no uniqueness claim is smuggled via self-citation. The self-citations present are normal scholarly practice and do not form a load-bearing circular chain.

Axiom & Free-Parameter Ledger

10 free parameters · 5 axioms · 0 invented entities

The paper does not invent new particles, forces, or entities. It uses established physics models (WIMPs, ALPs, pulsars, AGN) and established ML methods (BDTs, normalizing flows, weak supervision). The free parameters are either ML hyperparameters (selected by grid search), physics benchmark points (m_a, g_agamma), or experimental control variables (signal fraction, sample ratio). None are fitted to produce the central claim; they are scanned to characterize performance.

free parameters (10)
  • Signal fraction f in mixed sample = varied: 0.04-0.14 (pulsar-AGN), 0.04-0.14 (DM), 0.04-0.14 (ALP)
    The fraction of exotic/signal sources in the mixed sample M. Not fitted to data but scanned as a control parameter. Performance is reported as a function of f.
  • Background-to-mixture ratio |B|:|M| = 1:1 (balanced) and 2:1 (imbalanced)
    Relative sample sizes for B and M in the weakly supervised setup. Chosen to study the effect of training imbalance.
  • BDT number of estimators = 200 (pulsar-AGN, DM), 400 (ALP)
    Number of boosted trees. Selected by grid search on validation performance.
  • BDT max depth = 3
    Maximum tree depth. Selected by grid search.
  • BDT learning rate = not specified numerically
    Selected by grid search but exact value not reported.
  • Classifier score threshold = 0.5
    Fixed threshold for computing TPR/FPR. Authors state qualitative behavior is unchanged for other reasonable thresholds.
  • ALP mass m_a = 1 neV
    Fixed benchmark value for ALP-photon oscillation simulations. Choice discussed in Appendix B but not optimized.
  • ALP-photon coupling g_agamma = 50e-11 and 500e-11 GeV^-1
    Two benchmark coupling strengths chosen to represent weak and strong modulation regimes.
  • Normalizing flow: number of knots K = 10
    Spline knots for coupling transformations (Table 4). Architecture hyperparameter.
  • Normalizing flow: number of transformations = 8
    Number of sequential coupling transformations (Table 4). Architecture hyperparameter.
axioms (5)
  • domain assumption Same-background assumption: p_A(x) is distributed identically in background sample B and mixed sample M
    Stated in §2.2: 'Under the same-background assumption (the p_A component is distributed identically in B and M).' This is the foundational assumption of the BvM framework. All three case studies satisfy it by construction using simulated or matched data.
  • domain assumption The classifier is sufficiently expressive to learn the optimal decision boundary
    Implicit in §2.2: 'in the limit of sufficient training data and an expressive classifier, thresholds on the classifier score correspond to increasing evidence for an exotic origin.' BDTs with max depth 3 and 200-400 trees may not fully satisfy this in practice.
  • domain assumption Gamma-ray source spectra in fixed energy bins are sufficient features for classification
    §3.2: the baseline input is seven flux-band values from 4FGL. For ALP case, 50 energy bins are used (§5.2). The choice to exclude variability and positional information is motivated by model-agnostic considerations but limits achievable performance.
  • domain assumption The Jansson-Farrar GMF model adequately represents the Galactic magnetic field for ALP oscillation calculations
    §5.2: 'we use the Jansson-Farrar model for the regular GMF, with parameters as given in [54].' Updated models [55] and turbulent components [56] are deferred to future work.
  • domain assumption Simulated dark-matter subhalo spectra from N-body distributions are representative of real subhalo signals
    §4.2: subhalo population modeled using cosmological distributions from Aquarius simulations [4]. The detectability assessment uses fermipy and standard Fermi ScienceTools.

pith-pipeline@v1.1.0-glm · 24896 in / 3618 out tokens · 467435 ms · 2026-07-09T18:37:29.451902+00:00 · methodology

0 comments
read the original abstract

The $\gamma$-ray sky, as observed by the Fermi Large Area Telescope, contains a significant number of unassociated sources that may point to new astrophysical populations or more exotic phenomena. Machine-learning methods are widely used for source classification and searches for new physics, but most existing approaches rely on fully supervised training and therefore on explicit signal models. We explore weakly supervised classification as a less model-dependent strategy for analysing $\gamma$-ray source spectra. In a background-versus-mixture setup, classifiers are trained on samples with different signal admixtures rather than on fully labelled signal and background events. We study three representative scenarios: pulsar-active galactic nuclei separation as a controlled benchmark, the identification of dark-matter subhalos, and spectral irregularities induced by axion-photon oscillations. In each case we investigate the impact of signal fraction and sample composition on classification performance. Our results show that weak supervision can identify anomalous or signal-like subsets of data while reducing the reliance on detailed signal templates during training. In favourable cases, the method approaches the performance of fully supervised classifiers, while remaining applicable in situations where the signal model is uncertain or only partially specified. Weakly supervised learning therefore provides a complementary candidate-selection and anomaly-ranking strategy for $\gamma$-ray data analysis and searches for new phenomena.

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

55 extracted references · 55 canonical work pages · 49 internal anchors

  1. [1]

    The Large Area Telescope on the Fermi Gamma-ray Space Telescope Mission

    W.B. Atwood, A.A. Abdo, M. Ackermann, W. Althouse, B. Anderson et al.,The Large Area Telescope on the Fermi Gamma-Ray Space Telescope Mission,Astrophys. J.697(2009) 1071 [0902.1089]. [2]Fermi-LATcollaboration,F ermiLarge Area Telescope Fourth Source Catalog,Astrophys. J. Suppl.247(2020) 33 [1902.10045]. [3]Fermi-LATcollaboration,Fermi Large Area Telescope...

    work page internal anchor Pith review Pith/arXiv arXiv 2009

  2. [2]

    The Aquarius Project: the subhalos of galactic halos

    V. Springel, J. Wang, M. Vogelsberger, A. Ludlow, A. Jenkins, A. Helmi et al.,The Aquarius Project: the subhalos of galactic halos,Mon. Not. Roy. Astron. Soc.391(2008) 1685 [0809.0898]

    work page internal anchor Pith review Pith/arXiv arXiv 2008

  3. [3]

    Dark matter subhalos and the dwarf satellites of the Milky Way

    P. Madau, J. Diemand and M. Kuhlen,Dark matter subhalos and the dwarf satellites of the Milky Way,Astrophys. J.679(2008) 1260 [0802.2265]. [6]Fermi-LATcollaboration,Search for dark matter satellites using the fermi-lat,Astrophys. J. 747(2012) 121 [1201.2691]

    work page internal anchor Pith review Pith/arXiv arXiv 2008

  4. [4]

    Detecting Axion-Like Particles With Gamma Ray Telescopes

    D. Hooper and P.D. Serpico,Detecting Axion-Like Particles With Gamma Ray Telescopes, Phys. Rev. Lett.99(2007) 231102 [0706.3203]. [8]Fermi-LATcollaboration,Search for Spectral Irregularities due to Photon–Axionlike-Particle Oscillations with the Fermi Large Area Telescope,Phys. Rev. Lett.116(2016) 161101 [1603.06978]

    work page internal anchor Pith review Pith/arXiv arXiv 2007

  5. [5]

    Fermi's Sibyl: Mining the gamma-ray sky for dark matter subhaloes

    N. Mirabal, V. Frias-Martinez, T. Hassan and E. Frias-Martinez,Fermi’s Sibyl: Mining the gamma-ray sky for dark matter subhaloes,Mon. Not. Roy. Astron. Soc.424(2012) L64 [1205.4825]

    work page internal anchor Pith review Pith/arXiv arXiv 2012

  6. [6]

    Classification and Ranking of Fermi LAT Gamma-ray Sources from the 3FGL Catalog using Machine Learning Techniques

    P.M. Saz Parkinson, H. Xu, P.L.H. Yu, D. Salvetti, M. Marelli and A.D. Falcone,Classification and Ranking of Fermi LAT Gamma-ray Sources from the 3FGL Catalog using Machine Learning Techniques,Astrophys. J.820(2016) 8 [1602.00385]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  7. [7]

    Blazar Flaring Patterns (B-FlaP): Classifying Blazar Candidates of Uncertain type in the third Fermi-LAT catalog by Artificial Neural Networks

    G. Chiaro, D. Salvetti, G. La Mura, M. Giroletti, D.J. Thompson and D. Bastieri,Blazar flaring patterns (B-FlaP) classifying blazar candidate of uncertain type in the third Fermi-LAT catalogue by artificial neural networks,Mon. Not. Roy. Astron. Soc.462(2016) 3180 [1607.07822]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  8. [8]

    3FGLzoo. Classifying 3FGL Unassociated Fermi-LAT Gamma-ray Sources by Artificial Neural Networks

    D. Salvetti, G. Chiaro, G. La Mura and D.J. Thompson,3FGLzoo: classifying 3FGL unassociated Fermi-LATγ-ray sources by artificial neural networks,Mon. Not. Roy. Astron. Soc.470(2017) 1291 [1705.09832]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  9. [9]

    Classification of Fermi-LAT sources with deep learning using energy and time spectra

    T. Finke, M. Krämer and S. Manconi,Classification of Fermi-LAT sources with deep learning using energy and time spectra,Mon. Not. Roy. Astron. Soc.507(2021) 4061 [2012.05251]

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  10. [10]

    Classification of Fermi-LAT blazars with Bayesian neural networks

    A. Butter, T. Finke, F. Keil, M. Krämer and S. Manconi,Classification of Fermi-LAT blazars with Bayesian neural networks,JCAP04(2022) 023 [2112.01403]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  11. [11]

    3FGL Demographics Outside the Galactic Plane using Supervised Machine Learning: Pulsar and Dark Matter Subhalo Interpretations

    N. Mirabal, E. Charles, E.C. Ferrara, P.L. Gonthier, A.K. Harding, M.A. Sánchez-Conde et al., 3FGL Demographics Outside the Galactic Plane using Supervised Machine Learning: Pulsar and Dark Matter Subhalo Interpretations,Astrophys. J.825(2016) 69 [1605.00711]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  12. [12]

    Unidentified Gamma-ray Sources as Targets for Indirect Dark Matter Detection with the Fermi-Large Area Telescope

    J. Coronado-Blazquez, M.A. Sanchez-Conde, A. Dominguez, A. Aguirre-Santaella, M. Di Mauro, N. Mirabal et al.,Unidentified Gamma-ray Sources as Targets for Indirect Dark Matter Detection with the Fermi-Large Area Telescope,JCAP07(2019) 020 [1906.11896]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  13. [13]

    Machine-Learned Dark Matter Subhalo Candidates in the 4FGL-DR2: Search for the Perturber of the GD-1 Stream

    N. Mirabal and A. Bonaca,Machine-learned dark matter subhalo candidates in the 4FGL-DR2: search for the perturber of the GD-1 stream,JCAP11(2021) 033 [2105.12131]. – 25 –

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  14. [14]

    A search for dark matter among Fermi-LAT unidentified sources with systematic features in Machine Learning

    V. Gammaldi, B. Zaldívar, M.A. Sánchez-Conde and J. Coronado-Blázquez,A search for dark matter among Fermi-LAT unidentified sources with systematic features in machine learning, Mon. Not. Roy. Astron. Soc.520(2023) 1348 [2207.09307]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  15. [15]

    Search for dark matter subhalos among unassociated Fermi-LAT sources in presence of dataset shift

    A. Amerio, D. Malyshev, B. Zaldívar, V. Gammaldi and M.Á. Sánchez-Conde,Search for dark matter subhalos among unassociated fermi-lat sources in presence of dataset shift,2503.14584

    work page internal anchor Pith review Pith/arXiv arXiv

  16. [16]

    Searching for dark matter subhalos in the Fermi-LAT catalog with Bayesian neural networks

    A. Butter, M. Krämer, S. Manconi and K. Nippel,Searching for dark matter subhalos in the Fermi-LAT catalog with Bayesian neural networks,JCAP07(2023) 033 [2304.00032]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  17. [17]

    Hernández-González, I

    J. Hernández-González, I. Inza and J.A. Lozano,Weak supervision and other non-standard classification problems: A taxonomy,Pattern Recognit. Lett.69(2016) 49

    work page 2016

  18. [18]

    Zhou,A brief introduction to weakly supervised learning,National Science Review5 (2018) 44

    Z.-H. Zhou,A brief introduction to weakly supervised learning,National Science Review5 (2018) 44

    work page 2018

  19. [19]

    Classification without labels: Learning from mixed samples in high energy physics

    E.M. Metodiev, B. Nachman and J. Thaler,Classification without labels: Learning from mixed samples in high energy physics,JHEP10(2017) 174 [1708.02949]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  20. [20]

    Extending the Bump Hunt with Machine Learning

    J.H. Collins, K. Howe and B. Nachman,Extending the search for new resonances with machine learning,Phys. Rev. D99(2019) 014038 [1902.02634]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  21. [21]

    Karagiorgi, G

    G. Karagiorgi, G. Kasieczka, S. Kravitz, B. Nachman and D. Shih,Machine learning in the search for new fundamental physics,Nature Rev. Phys.4(2022) 399

    work page 2022

  22. [22]

    Machine Learning for Anomaly Detection in Particle Physics

    V. Belis, P. Odagiu and T.K. Aarrestad,Machine learning for anomaly detection in particle physics,Rev. Phys.12(2024) 100091 [2312.14190]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  23. [23]

    Via Machinae: Searching for Stellar Streams using Unsupervised Machine Learning

    D. Shih, M.R. Buckley, L. Necib and J. Tamanas,via machinae: Searching for stellar streams using unsupervised machine learning,Mon. Not. Roy. Astron. Soc.509(2021) 5992 [2104.12789]

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  24. [24]

    Via Machinae 2.0: Full-Sky, Model-Agnostic Search for Stellar Streams in Gaia DR2

    D. Shih, M.R. Buckley and L. Necib,Via Machinae 2.0: Full-sky, model-agnostic search for stellar streams in Gaia DR2,Mon. Not. Roy. Astron. Soc.529(2024) 4745 [2303.01529]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  25. [25]

    Weakly-Supervised Anomaly Detection in the Milky Way

    M. Pettee, S. Thanvantri, B. Nachman, D. Shih, M.R. Buckley and J.H. Collins,Weakly supervised anomaly detection in the Milky Way,Mon. Not. Roy. Astron. Soc.527(2023) 8459 [2305.03761]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  26. [26]

    Via Machinae 3.0: A search for stellar streams in Gaia with the CATHODE algorithm

    A. Hallin, D. Shih, C. Krause and M.R. Buckley,Via Machinae 3.0: A search for stellar streams in Gaia with the CATHODE algorithm,2509.08064

    work page internal anchor Pith review Pith/arXiv arXiv

  27. [27]

    Characterization and classification of $\gamma$-ray bursts from blazars

    M. Cerruti,Characterization and classification ofγ-ray bursts from blazars,Astron. Astrophys. 698(2025) A101 [2410.21974]

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  28. [28]

    Friedman,Greedy function approximation: A gradient boosting machine.,Annals of Statistics29(2001) 1189

    J.H. Friedman,Greedy function approximation: A gradient boosting machine.,Annals of Statistics29(2001) 1189

    work page 2001

  29. [29]

    Back To The Roots: Tree-Based Algorithms for Weakly Supervised Anomaly Detection

    T. Finke, M. Hein, G. Kasieczka, M. Krämer, A. Mück, P. Prangchaikul et al.,Tree-based algorithms for weakly supervised anomaly detection,Phys. Rev. D109(2024) 034033 [2309.13111]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  30. [30]

    Why do tree-based models still outperform deep learning on tabular data?

    L. Grinsztajn, E. Oyallon and G. Varoquaux,Why do tree-based models still outperform deep learning on tabular data?,ArXiv(2022) [2207.08815]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  31. [31]

    Classifying Anomalies THrough Outer Density Estimation (CATHODE)

    A. Hallin, J. Isaacson, G. Kasieczka, C. Krause, B. Nachman, T. Quadfasel et al.,Classifying anomalies through outer density estimation,Phys. Rev. D106(2022) 055006 [2109.00546]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  32. [32]

    Variational Inference with Normalizing Flows

    D.J. Rezende and S. Mohamed,Variational inference with normalizing flows,ArXiv(2015) [1505.05770]

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  33. [33]

    Density estimation using Real NVP

    L. Dinh, J.N. Sohl-Dickstein and S. Bengio,Density estimation using real nvp,ArXiv(2016) [1605.08803]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  34. [34]

    Neural Spline Flows

    C. Durkan, A. Bekasov, I. Murray and G. Papamakarios,Neural spline flows,1906.04032. – 26 –

    work page internal anchor Pith review Pith/arXiv arXiv 1906

  35. [35]

    PPPC 4 DM ID: A Poor Particle Physicist Cookbook for Dark Matter Indirect Detection

    M. Cirelli, G. Corcella, A. Hektor, G. Hutsi, M. Kadastik, P. Panci et al.,PPPC 4 DM ID: A Poor Particle Physicist Cookbook for Dark Matter Indirect Detection,JCAP03(2011) 051 [1012.4515]

    work page internal anchor Pith review Pith/arXiv arXiv 2011

  36. [36]

    M.J. Wood, R. Caputo, E. Charles, M. di Mauro, J.D. Magill and J.S. Perkins,Fermipy: An open-source python package for analysis of fermi-lat data,arXiv: Instrumentation and Methods for Astrophysics(2017)

    work page 2017

  37. [37]

    The landscape of QCD axion models

    L. Di Luzio, M. Giannotti, E. Nardi and L. Visinelli,The landscape of QCD axion models, Phys. Rept.870(2020) 1 [2003.01100]

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  38. [38]

    Axion Dark Matter: What is it and Why Now?

    F. Chadha-Day, J. Ellis and D.J.E. Marsh,Axion dark matter: What is it and why now?,Sci. Adv.8(2022) abj3618 [2105.01406]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  39. [39]

    Cosmology of axion dark matter

    C.A.J. O’Hare,Cosmology of axion dark matter,PoSCOSMICWISPers(2024) 040 [2403.17697]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  40. [40]

    Gamma-ray spectral modulations of Galactic pulsars caused by photon-ALPs mixing

    J. Majumdar, F. Calore and D. Horns,Search for gamma-ray spectral modulations in galactic pulsars,Journal of Cosmology and Astroparticle Physics2018(2018) 048–048 [1801.08813]

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  41. [41]

    Z.-Q. Xia, C. Zhang, Y.-F. Liang, L. Feng, Q. Yuan, Y.-Z. Fan et al.,Searching for spectral oscillations due to photon-axionlike particle conversion using the Fermi-LAT observations of bright supernova remnants,Phys. Rev. D97(2018) 063003 [1801.01646]

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  42. [42]

    Constraints on axion-like particle properties with very high energy gamma-ray observations of Galactic sources

    Y.-F. Liang, C. Zhang, Z.-Q. Xia, L. Feng, Q. Yuan and Y.-Z. Fan,Constraints on axion-like particle properties with TeV gamma-ray observations of Galactic sources,JCAP06(2019) 042 [1804.07186]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  43. [43]

    Reconciling hints on axion-like-particles from high-energy gamma rays with stellar bounds

    G.A. Pallathadka et al.,Reconciling hints on axion-like-particles from high-energy gamma rays with stellar bounds,JCAP11(2021) 036 [2008.08100]

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  44. [44]

    Axion-Like Particle Searches with IACTs

    I. Batkovic, A. De Angelis, M. Doro and M. Manganaro,Axion-like particle searches with iacts, Universe7(2021) 185 [2106.03424]

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  45. [45]

    Astrophysical limits on very light axion-like particles from Chandra grating spectroscopy of NGC 1275

    C.S. Reynolds, M.C.D. Marsh, H.R. Russell, A.C. Fabian, R. Smith, F. Tombesi et al., Astrophysical limits on very light axion-like particles from Chandra grating spectroscopy of NGC 1275,Astrophys. J.890(2020) 59 [1907.05475]

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  46. [46]

    Detecting ALP wiggles at TeV energies

    M. Kachelriess and J. Tjemsland,Detecting ALP wiggles at TeV energies,JCAP01(2024) 044 [2305.03604]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  47. [47]

    Projected sensitivity of CTAO to axion-like particles from blazars with a machine learning approach

    F. Schiavone, L. Di Venere and F. Giordano,Projected sensitivity of CTAO to axion-like particles from blazars with a machine learning approach,2512.19259

    work page internal anchor Pith review Pith/arXiv arXiv

  48. [48]

    Stochastic conversions of TeV photons into axion-like particles in extragalactic magnetic fields

    A. Mirizzi and D. Montanino,Stochastic conversions of tev photons into axion-like particles in extragalactic magnetic fields,Journal of Cosmology and Astroparticle Physics2009(2017) 004–004 [0911.0015]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  49. [49]

    Practical Modeling of Large-Scale Galactic Magnetic Fields: Status and Prospects

    T.R. Jaffe,Practical Modeling of Large-Scale Galactic Magnetic Fields: Status and Prospects, Galaxies7(2019) 52 [1904.12689]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  50. [50]

    A New Model of the Galactic Magnetic Field

    R. Jansson and G.R. Farrar,A New Model of the Galactic Magnetic Field,ApJ, Volume 757, Issue 1, article id. 14, 13 pp. (2012)(2012) [1204.3662]

    work page internal anchor Pith review Pith/arXiv arXiv 2012

  51. [51]

    The Coherent Magnetic Field of the Milky Way

    M. Unger and G.R. Farrar,The Coherent Magnetic Field of the Milky Way,Astrophys. J.970 (2024) 95 [2311.12120]

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  52. [52]

    Turbulent axion-photon conversions in the Milky-Way

    P. Carenza, C. Evoli, M. Giannotti, A. Mirizzi and D. Montanino,Turbulent axion-photon conversions in the Milky Way,Phys. Rev. D104(2021) 023003 [2104.13935]

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  53. [53]

    gammaALPs: An open-source python package for computing photon-axion-like-particle oscillations in astrophysical environments

    M. Meyer, J. Davies and J. Kuhlmann,gammaalps: An open-source python package for computing photon-axion-like-particle oscillations in astrophysical environments,2108.02061. [58]Fermi-LATcollaboration,The Third Fermi Large Area Telescope Catalog of Gamma-Ray Pulsars,Astrophys. J.958(2023) 191 [2307.11132]. – 27 –

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  54. [54]

    Bhattacharjee, C

    P. Bhattacharjee, C. Eckner, G. Zaharijas, G. Kluge and G. D’Amico,Probing the Parameter Space of Axion-Like Particles Using Simulation-Based Inference,2509.20578

    work page arXiv

  55. [55]

    Cubic-Spline Flows

    C. Durkan, A. Bekasov, I. Murray and G. Papamakarios,Cubic-spline flows,ArXiv(2019) [1906.02145]. – 28 –

    work page internal anchor Pith review Pith/arXiv arXiv 2019