Pith. sign in

REVIEW 3 major objections 5 minor 51 references

Machine learning methods for spectroscopic information recovery under ultrafast photon pileup

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that 3D convolutional networks can recover signal fractions and photon multiplicities from spatial energy deposition maps when ultrafast pileup destroys time information.

desk verdict Genuinely new problem framing for spatial pileup recovery, but the reported accuracies are likely optimistic because the synthetic pileup preserves depth information real pixelated CZT cannot resolve. read the letter →

arxiv 2608.10143 v1 pith:EQZYSC7V submitted 2026-08-10 physics.ins-det

Jayson R. Vavrek , Thomas D. MacDonald , Yue Shi Lai This is my paper · ORCID
classification physics.ins-det PACS 29.40.Wk07.05.Mh
keywords photonpileupspectroscopicinformationrecoveryposition-sensitivedetectors3DconvolutionalneuralnetworksignalfractionestimationmultiplicityCdZnTedetectormachinelearninginnucleardetection
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper sets out to recover spectroscopic information that is normally lost when many photons hit a detector within a single ultrafast pulse. Instead of separating photons in time, it treats the piled-up event as a three-dimensional map of energy deposition and asks whether neural networks can read the source composition and photon count from that map. In the synthetic signal-fraction problem, a 3D convolutional network reconstructs the fraction of energy coming from a monoenergetic signal with 14.5% RMSE, outperforming fully-connected (18.8%), point-cloud (16.8%), and classical depth-ratio (28–41%) baselines. In the multiplicity problem on synthetically-piled-up real Cs-137 data, the CNN recovers the number of photons with 0.61 RMSE, beating total-energy and voxel-count baselines. A sympathetic reading of these results is that spatial pileup patterns contain enough structure to re-enable spectroscopy in fast-pulsed source applications.

What carries the argument

The central object is the three-dimensional energy deposition heatmap: the voxelized, timestamp-free record of every energy deposit left in the detector by a pileup pulse. The load-bearing mechanism is a 3D convolutional neural network that slides kernels over this heatmap to learn spatial correlations — such as the depth-dependent stopping of higher-energy photons and Compton-scatter correlations between separated voxels — and regresses the target quantity. The paper compares this to a flattened fully-connected network, to PointNet++, a point-cloud network that ingests (x, y, z, E) hit lists, and to classical baselines that reduce the heatmap to a single front-to-back depth ratio. The heatmap representation is what lets the CNN exploit physics that the depth-ratio models cannot see.

What would settle it

A direct test would run the same 3D CNN on data from a pulsed source in which the true per-pulse multiplicity is known from a slow timestamp-resolved readout or a second detector; if the network's RMSE on multiplicity or signal fraction moves well above the reported 0.61 and 14.5% once real charge-sharing and depth ambiguity are present, the synthetic-pileup model is too optimistic.

Watch

Extended reading notes

Core claim

The central claim is that the spatial energy deposition pattern left by a pileup pulse encodes enough information to reconstruct spectroscopic quantities, and that machine-learning models can extract that information where classical depth-only methods cannot. The paper demonstrates this for two representative problems: estimating the signal energy deposition fraction f in a mixture of monoenergetic signal and bremsstrahlung background, and estimating the photon multiplicity K of a pure source. On synthetic data with up to twenty piled-up photons, the 3D CNN reaches 14.5% RMSE on f, the FCNN 18.8%, and PointNet++ 16.8%, while the best classical depth-threshold models reach only 28% and 41%. On synthetically-piled-up real Cs-137 detector data, the same CNN architecture recovers K with 0.61 RMSE (0.44 mean absolute error), compared to 1.56 for the total-energy baseline and 0.74 for the voxel-count baseline. The paper's conclusion is that these proof-of-concept methods can be adapted to mission-specific active-interrogation detectors.

Load-bearing premise

The load-bearing premise is that a real pileup readout behaves like the sum of independent single-photon interaction records after Gaussian energy smearing and coarse spatial binning; if real charge sharing, dead pixels, efficiency variations, or depth-of-interaction ambiguity break that equivalence, the reported accuracies would not transfer to the field.

Editorial extensions

If this is right

  • Pulsed-beam detectors can continue to perform spectroscopy at fluxes where individual photons cannot be time-resolved, recovering the signal energy fraction to about 14.5% RMSE in the tested synthetic setting.
  • Photon multiplicities of pure sources can be estimated with sub-photon error from piled-up real Cs-137 data, preserving counting information that ordinary readouts would discard.
  • The classical depth-threshold and likelihood baselines saturate at 28–41% RMSE and cannot improve with more data, whereas the machine-learning models improve as training sets grow.
  • Because the 3D CNN outperforms flattened and point-cloud networks on these problems, preserving the spatial structure of the heatmap is likely important, and further CNN tuning may narrow the gap further.
  • The same training pipeline can be adapted to other detector geometries and other reconstructed quantities, such as separate signal and background multiplicities, rather than only the two scalars demonstrated.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The reported RMSEs should be read as lower bounds for real-detector performance, since the synthetic and synthetically-piled-up data omit charge sharing, dead pixels, efficiency variations, and depth-of-interaction ambiguity; adding those effects to the simulator would quantify the expected field degradation.
  • If the heatmap carries enough information, the same networks could be trained to reconstruct full deposited-energy spectra or joint signal/background counts rather than scalars, which would make the method directly useful for isotope identification.
  • A practical deployment path would fine-tune a network trained on synthetic pileup with a small set of measured piled-up events, since unmodeled detector non-idealities are the most likely cause of a performance gap.
  • Reorienting a pixelated detector so its depth segmentation lies along the beam direction could trade transverse fidelity for more reliable depth information, which the paper suggests may help networks that depend on depth patterns.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper addresses spectroscopic readout under ultrafast pileup by replacing temporal separation with spatial energy-deposition patterns in position-sensitive detectors. It defines two tasks: signal energy fraction recovery (monoenergetic signal against a smooth bremsstrahlung background) and photon multiplicity recovery. Synthetic Geant4 CZT data are voxelized and summed to form piled-up pulses; a 3D CNN, a fully connected network, PointNet++, and two depth-threshold classical baselines are compared, with the 3D CNN achieving 14.5% RMSE on the signal-fraction task. A second demonstration uses measured Cs-137 single-photon event records that are synthetically piled up to train a CNN that recovers multiplicity with RMSE 0.61, outperforming total-energy and voxel-count baselines. The paper also includes an event-sequence counting analysis in Appendix A and an MC-dropout uncertainty-quantification study in Appendix B.

Significance. If the headline results hold under realistic detector readout, the paper would be a useful contribution: it gives a tractable route to spectroscopy under extreme pileup, defines concrete benchmark tasks with multiple baselines, and includes an honest limitations discussion. Strengths include the direct comparison of several ML architectures, the use of measured single-photon records for the multiplicity demonstration, and the explicit combinatorial complexity argument in Appendix A. The principal weakness is that the synthetic pileup model may overstate the depth information available from real pixelated CZT detectors; this is acknowledged in Section V but not quantified, so the practical significance for real detectors remains to be established.

major comments (3)
  1. [§III-F and §V] The piled-up pulses in Section III-F are constructed by discretizing each single-photon event record into voxels and then summing the records, so two simultaneous interactions in the same anode pixel at different depths remain two distinct depth voxels. The fourth paragraph of Section V states that real pixelated-anode CZT detectors cannot readily distinguish such events, making the results 'likely optimistic.' This mismatch is load-bearing because the 0.61 multiplicity RMSE and the 14.5% signal-fraction RMSE both rely on depth information being present in the input. The manuscript should quantify the effect by adding an ablation in which same-pixel coincident interactions are merged into one readout (for example, summing energies and collapsing depth to a centroid or removing the depth coordinate entirely) and reporting the resulting RMSEs. If the degradation is substantial, the central claims about re-enabling spectroscopy with these spatial patterns need to be tempered.
  2. [§IV-A and §IV-B (Table II, Fig. 10)] All headline RMSE values are single-run point estimates without error bars, multiple training seeds, or confidence intervals. The claimed ordering CNN (14.5%) < PointNet++ (16.8%) < FCNN (18.8%) could plausibly be affected by initialization and training stochasticity. Reporting mean and standard deviation over at least five seeds, or bootstrap confidence intervals on the validation set, would make the architecture comparisons and the claimed superiority over the classical baselines statistically meaningful.
  3. [§III-A and §IV-A] The models are evaluated on validation data drawn from the same uniform K_s, K_b in [0,10] distribution used for training, so the reported RMSEs are interpolation results within a synthetic design. The abstract's practical claim that the methods can help 're-enable spectroscopic analyses' in active-interrogation applications requires some evidence of transfer beyond this training distribution. A minimal out-of-distribution test, such as a different multiplicity range, a different signal energy, or a different background endpoint E_max, would substantially strengthen the generalization claim; without it, the practical reach of the results is unclear.
minor comments (5)
  1. [Fig. 1 and §III-A] Figure 1 is described as a large block of germanium while the simulations and the experimental demonstration use CZT; the caption should state explicitly that the figure is a schematic illustration, not a model of the simulated detector.
  2. [§IV-A1] The statement that the CNN 'fails to predict any signal fractions lower than f-hat approximately 0.1' deserves a brief explanation of the consequence for the reported 14.5% RMSE, since the response matrix shows a systematic rather than purely random deviation at low true fractions.
  3. [§III-D] PointNet++ inputs are zero-padded to 100 spatial hits, but the energy coordinate E is also zero for padded entries; the zero-padded points with (x,y,z) = (0,0,0) and E = 0 may act as a spurious cluster at the origin. The authors should clarify whether padded points are masked in the network or whether their effect was tested.
  4. [§III-F and §IV-B] The real-data demonstration uses synthetically piled-up isolated single-photon records; this is stated clearly, but the caption of Fig. 9 and the text should also remind readers that real pileup includes electronics dead time, charge sharing, and possible event-loss mechanisms not present in the synthetic pileup.
  5. [Appendix B] The phrase 'applying dropout after every network unit in each layer' is ambiguous; the authors should specify whether this refers to dropout after every convolutional and dense layer (as opposed to only before the final dense layer) and how that differs from the architecture used in the main results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the ML predictions are evaluated against independent held-out targets, and no fitted parameter is disguised as a prediction.

full rationale

The paper's central results are supervised regression tasks: the CNN, FCNN, and PointNet++ are trained on synthetic piled-up heatmaps whose target labels (signal energy fraction f or multiplicity K) are set by construction of the training samples, and performance is reported on disjoint validation sets. No equation in the paper defines the predicted quantity in terms of the model output, nor is any fitted parameter renamed as a prediction. The classical baselines (depth-ratio linear regression, depth-ratio maximum likelihood, total-energy division, and voxel-count division) are also external estimators evaluated on the same held-out data, so the ML improvements are not forced by the training setup. The only self-citations are speculative future-work remarks (e.g., coupling to Ref. [41]) and do not carry the derivation. The acknowledged limitation in Section V—that real pixelated CZT detectors cannot readily resolve simultaneous different-depth interactions in the same pixel, so results are likely optimistic—is a realism caveat about the synthetic forward model, not a circularity: it does not make the reported predictions equivalent to their inputs by definition. The derivation chain is therefore self-contained as a machine-learning benchmark, even though real-detector transfer remains untested.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim depends on the fidelity of the simulation and synthetic pileup construction, on the choice of training distribution, and on the adequacy of the chosen baselines. No new physical entities are introduced. The free parameters are modeling choices and classical-method hyperparameters that affect the reported quantitative results.

free parameters (7)
  • depth threshold d0 (linear regression) = bin 5 of 50
    Chosen to minimize validation RMSE; the classical model's predictions depend strongly on this choice. Section IV-A4.
  • depth threshold d0 (maximum likelihood) = bin 34 (instructive) or bin 48 (RMSE-minimizing)
    Chosen by RMSE scan; the model collapses to a constant prediction at the RMSE-minimizing value. Section IV-A4.
  • number of histogram bins n_bins (maximum likelihood) = 50
    Chosen by hand for the R_d0 vs f histogram; affects the likelihood model resolution. Section III-E.
  • background endpoint energy E_max = 2.5 MeV
    Sets the bremsstrahlung background energy distribution p(E) ~ (E_max/E) - 1; influences the spatial patterns the models must learn. Section III-A.
  • sampling bounds for K_s and K_b = uniform in [0,10] each (up to 20 total)
    Determines the training distribution of signal fractions and multiplicities; the paper notes low-f pulses are underrepresented. Section III-A.
  • voxelization grid = 20x20x50 in x,y,z
    Spatial discretization of the detector; defines the input representation for CNN and FCNN and limits spatial information. Section III-A.
  • PointNet++ two-pass resampling = inverse-frequency acceptance-rejection over 100 f-bins
    A data weighting scheme chosen to equalize the signal fraction distribution; affects PointNet++ performance and comparisons. Section III-D.
assumptions (5)
  • domain assumption Geant4 with Livermore physics lists accurately models photon interactions in CZT for this study.
    The synthetic data are generated with this simulation; all learned patterns inherit this assumption. Section III-A.
  • domain assumption Piled-up pulses are formed by summing independent single-photon interaction records with linear energy readout, ignoring charge sharing, dead pixels, efficiency variations, and depth-of-interaction ambiguities.
    This is the core surrogate model for pileup. The paper explicitly lists these omissions as limitations in Section V.
  • domain assumption The training/validation distribution (uniform K_s,K_b) is representative of the target operational distribution.
    RMSE is evaluated on data drawn from the same distribution used for training; real deployments may have different multiplicity and fraction priors. Section III-A and IV.
  • domain assumption The depth-ratio classical methods are an adequate baseline for comparing ML vs non-ML information recovery.
    The paper intentionally uses only depth information for classical methods and notes maximum likelihood is computationally infeasible, so the baseline is a lower bound on classical performance. Sections III-E and V.
  • domain assumption Single-photon events from the measured H3D M400i Cs-137 dataset, when synthetically combined, represent true piled-up multi-photon events.
    The experimental multiplicity demonstration relies on this; no measured true pileup data are used. Section III-F.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Machine learning methods for spectroscopic information recovery under ultrafast photon pileup." pith.science (2026). https://pith.science/paper/EQZYSC7V

@misc{pith2026260810143,
  author       = {Pith},
  title        = {Pith review of: Machine learning methods for spectroscopic information recovery under ultrafast photon pileup},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EQZYSC7V}},
  note         = {Machine review of arXiv:2608.10143}
}
abstract

We present methods for recovering spectroscopic information from multiple concurrent photon interactions that would normally be lost due to pulse pileup. In particular, we focus on machine learning methods to recover information based on spatial (rather than temporal) energy deposition patterns in position-sensitive detectors. We construct two representative problems, namely (1) recovering the fraction of total energy deposition stemming from a monoenergetic signal vs. a smooth background; and (2) recovering the signal multiplicity, i.e., the number of interacting photons, in a pure-source-term example. In the signal fraction recovery problem, we use 3D convolutional neural networks (CNNs), fully-connected neural networks (FCNNs), a network based on the PointNet++ architecture, and two non-machine-learning methods to estimate the signal fraction in synthetic data when up to 20 total piled-up photons are present. The CNN, FCNN, and PointNet++ models reconstruct the signal energy deposition fractions with root mean square errors (RMSEs) of $14.5\%$, $18.8\%$, and $16.8\%$ given training datasets that fit in-core, while the classical methods perform poorly and will not improve with additional training data. In the multiplicity recovery problem, we demonstrate that, when trained with synthetically-piled-up real Cs-137 data, the 3D CNN architecture can recover the multiplicity with sub-photon RMSE, outperforming non-ML baselines. These methods can be adapted to future, more specific photon active interrogation applications, helping to re-enable spectroscopic analyses in those domains.

Figures

Figures reproduced from arXiv: 2608.10143 by the authors.

Figure 1
Figure 1. Toy model overview of the ultrafast pileup and information recovery problem, using NRF signal photons and [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Results for the 3D convolutional neural network. Top left: training data response matrix. Top right: validation data [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Results for the best performing single- and two-layer fully connected neural networks. (Top row) [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Results for the PointNet++ model. (a): training response matrix, (b): training RMSE distribution over true signal [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Training and validation RMSE as the function of [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Results for the classical linear regression model. (a) RMSE between predicted [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Results for the classical likelihood model with threshold [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Front-to-total energy deposition ratios, [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 10
Figure 10. Figure 10: RMSE vs. true multiplicity K for the CNN and the two baseline (non-ML) methods. Fig. A.1: Number of interaction sequences given N hits. Although N takes whole-number values, continuous curves are drawn for visual clarity. networks and several methods for recalibration…
Figure 9
Figure 9. Figure 9: Real-data multiplicity recovery results. Top: Poisson [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 43 canonical work pages

  1. [1]

    The BELLA Center hundred terawatt laser system for photon sources and user experiments,

    T. Ostermayr, H.-E. Tsai, R. Ettelbrick, L. Fan-Chiang, R. Jacob, A. Laut, O. Zhou, J. van Tilborg, F. Isono, S. K. Barberet al., “The BELLA Center hundred terawatt laser system for photon sources and user experiments,” inAPS Division of Plasma Physics Meeting Abstracts, vol. 2020, 2020, pp. BO04–002

    work page 2020

  2. [2]

    Development of the MeV Thomson-scattered gamma ray source using laser plasma accelerators at the BELLA Center,

    Q. Chen, R. Jacob, J. van Tilborg, A. Gonsalves, K. Nakamura, C. Schroeder, E. Esarey, and C. Geddes, “Development of the MeV Thomson-scattered gamma ray source using laser plasma accelerators at the BELLA Center,” inCompact Radiation Sources from EUV to Gamma-rays: Development and Applications. SPIE, 2023, p. PC125820C

    work page 2023

  3. [3]

    Principles and applications of x-ray light sources driven by laser wakefield acceleration,

    F. Albert, “Principles and applications of x-ray light sources driven by laser wakefield acceleration,”Physics of Plasmas, vol. 30, no. 5, 2023

    work page 2023

  4. [4]

    Impact of monoenergetic photon sources on nonproliferation applications final report,

    C. Geddes, B. Ludewigt, J. Valentine, B. Quiter, M.-A. Descalle, G. War- ren, M. Kinlaw, S. Thompson, D. Chichester, C. Milleret al., “Impact of monoenergetic photon sources on nonproliferation applications final report,” Idaho National Laboratory (INL), Idaho Falls, ID (United States), Tech. Rep., 2017

    work page 2017

  5. [5]

    Resonance fluorescence in nuclei,

    F. R. Metzger, “Resonance fluorescence in nuclei,”Prog Nuc Phys, vol. 7, p. 54, 1959

    work page 1959

  6. [6]

    Investigation of nuclear structure by resonance fluorescence scattering,

    U. Kneissl, H. Pitz, and A. Zilges, “Investigation of nuclear structure by resonance fluorescence scattering,”Progress in Particle and Nuclear Physics, vol. 37, pp. 349–433, 1996

    work page 1996

  7. [7]

    Transmission-based detection of nuclides with nuclear resonance fluorescence using a quasimonoener- getic photon source,

    C. Hagmann, J. Hall, M. Johnson, D. McNabb, J. Kelley, C. Huibregtse, E. Kwan, G. Rusev, and A. Tonchev, “Transmission-based detection of nuclides with nuclear resonance fluorescence using a quasimonoener- getic photon source,”Journal of Applied Physics, vol. 106, no. 8, 2009

    work page 2009

  8. [8]

    Nuclear resonance fluorescence for nuclear materials assay,

    B. J. Quiter, “Nuclear resonance fluorescence for nuclear materials assay,” Ph.D. dissertation, University of California, Berkeley, 2010

    work page 2010

Show all 51 references
  1. [9]

    Physical cryptographic verification of nuclear warheads,

    R. S. Kemp, A. Danagoulian, R. R. Macdonald, and J. R. Vavrek, “Physical cryptographic verification of nuclear warheads,”Proceedings of the National Academy of Sciences, vol. 113, no. 31, pp. 8618–8623, 2016

    2016

  2. [10]

    Experimental demonstration of an isotope-sensitive warhead verification technique using nuclear resonance fluorescence,

    J. R. Vavrek, B. S. Henderson, and A. Danagoulian, “Experimental demonstration of an isotope-sensitive warhead verification technique using nuclear resonance fluorescence,”Proceedings of the National Academy of Sciences, vol. 115, no. 17, pp. 4363–4368, 2018

    2018

  3. [11]

    Rapid interrogation of special nuclear materials by combining scattering and transmission nuclear resonance fluorescence spectroscopy,

    H.-Y . Lan, T. Song, J.-L. Zhang, J.-L. Zhou, and W. Luo, “Rapid interrogation of special nuclear materials by combining scattering and transmission nuclear resonance fluorescence spectroscopy,”Nuclear Sci- ence and Techniques, vol. 32, no. 8, p. 84, 2021

    2021

  4. [12]

    Isotope-sensitive imaging of special nuclear materials using computer tomography based on scattering nuclear resonance fluorescence,

    H. Lan, T. Song, Z. Luo, J. Zhou, Z. Zhu, and W. Luo, “Isotope-sensitive imaging of special nuclear materials using computer tomography based on scattering nuclear resonance fluorescence,”Physical Review Applied, vol. 16, no. 5, p. 054048, 2021

    2021

  5. [13]

    Review of high energy x- ray computed tomography for non-destructive dimensional metrology of large metallic advanced manufactured components,

    W. Sun, D. R. Symes, C. M. Brenner, M. B ¨ohnel, S. Brown, M. N. Mavrogordato, I. Sinclair, and M. Salamon, “Review of high energy x- ray computed tomography for non-destructive dimensional metrology of large metallic advanced manufactured components,”Reports on Progress in Ph...

    2022

  6. [14]

    The next genera- tion of MeV energy x-ray sources for use in the inspection of additively manufactured parts for industry,

    C. Thornton, S. Karimi, S. Glenn, W. Brown, N. Draganic, M. Skeate, M. Ferrucci, Q. Chen, R. Jacob, K. Nakamuraet al., “The next genera- tion of MeV energy x-ray sources for use in the inspection of additively manufactured parts for industry,”arXiv preprint arXiv:2404.09270, 2024

    2024 arXiv

  7. [15]

    Fundamental limitations of dual energy x-ray scanners for cargo content atomic number discrimination,

    P. Lalor and A. Danagoulian, “Fundamental limitations of dual energy x-ray scanners for cargo content atomic number discrimination,”Applied Radiation and Isotopes, vol. 206, p. 111201, 2024

    2024

  8. [16]

    Aγ-ray tracking algorithm for the GRETA spectrometer,

    G. Schmid, M. Deleplanque, I. Lee, F. Stephens, K. Vetter, R. Clark, R. Diamond, P. Fallon, A. Macchiavelli, and R. MacLeod, “Aγ-ray tracking algorithm for the GRETA spectrometer,”Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrome- ters, Det...

    1999

  9. [17]

    Bayesian reconstruction of photon interaction sequences for high-resolution PET detectors,

    G. Pratx and C. S. Levin, “Bayesian reconstruction of photon interaction sequences for high-resolution PET detectors,”Physics in Medicine & Biology, vol. 54, no. 17, p. 5073, 2009

    2009

  10. [18]

    Pileup mitigation with machine learning (PUMML),

    P. T. Komiske, E. M. Metodiev, B. Nachman, and M. D. Schwartz, “Pileup mitigation with machine learning (PUMML),”Journal of High Energy Physics, vol. 2017, no. 12, p. 51, 2017

    2017

  11. [19]

    A deep learning approach to correctly identify the sequence of coincidences in cross-strip CZT detectors,

    N. Nasiri and S. Abbaszadeh, “A deep learning approach to correctly identify the sequence of coincidences in cross-strip CZT detectors,” in Medical Imaging 2021: Physics of Medical Imaging, vol. 11595. SPIE, 2021, pp. 1011–1018

    2021

  12. [20]

    Experimental evaluation of convolutional neural network-based inter-crystal scattering recovery for high-resolution PET detectors,

    S. Lee and J. S. Lee, “Experimental evaluation of convolutional neural network-based inter-crystal scattering recovery for high-resolution PET detectors,”Physics in Medicine & Biology, vol. 68, no. 9, p. 095017, 2023

    2023

  13. [21]

    Classical and machine learning methods for event reconstruction in NeuLAND,

    J. Mayer, K. Boretzky, C. Douma, E. Hoemann, A. Zilgeset al., “Classical and machine learning methods for event reconstruction in NeuLAND,”Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equip- ment, vol. 101...

    2021

  14. [22]

    Low activity tritium detection in CCDs using deep learning techniques,

    E. Rofors, R. Heller, R. Cooper, J. Estrada, G. Moroni, B. Nachman, and K. Spears, “Low activity tritium detection in CCDs using deep learning techniques,”IEEE Transactions on Nuclear Science, vol. 73, no. 1, pp. 205–215, 2025

    2025

  15. [23]

    Geant4—a simulation toolkit,

    S. Agostinelli, J. Allison, K. Amako, J. Apostolakis, H. Araujo, P. Arce, M. Asai, D. Axen, S. Banerjee, G. Barrandet al., “Geant4—a simulation toolkit,”Nuclear Instruments and Methods in Physics Research Section 13 Fig. B.2: MC-dropout UQ results for the signal fraction recov...

    2003

  16. [24]

    Imaging and radiography with nuclear resonance fluorescence and effective-Z (EZ-3D™) determination; SNM detection using prompt neutrons from photon induced fission,

    W. Bertozzi, R. Hasty, A. Klimenko, S. E. Korbly, R. J. Ledoux, and W. Park, “Imaging and radiography with nuclear resonance fluorescence and effective-Z (EZ-3D™) determination; SNM detection using prompt neutrons from photon induced fission,” inAIP Conference Proceedings, vol...

    2009

  17. [25]

    M400 base specifications,

    H3D, Inc., “M400 base specifications,” retrieved February 18, 2025 from https://h3dgamma.com/M400Specs.pdf

    2025

  18. [26]

    Large-volume cad- mium zinc telluride modules for safeguards verification of unirradiated nuclear material,

    Y . Dodane, C. Schoch, S. Markin, and A. Lebrun, “Large-volume cad- mium zinc telluride modules for safeguards verification of unirradiated nuclear material,”Institute of Nuclear Materials Management, 2023

    2023

  19. [27]

    Many-electron singularity in x-ray photoe- mission and x-ray line spectra from metals,

    S. Doniach and M. Sunjic, “Many-electron singularity in x-ray photoe- mission and x-ray line spectra from metals,”Journal of Physics C: Solid State Physics, vol. 3, no. 2, p. 285, 1970

    1970

  20. [28]

    Gamma-ray peak shapes from cadmium zinc telluride detectors,

    M. Namboodiri, A. Lavietes, and J. McQuaid, “Gamma-ray peak shapes from cadmium zinc telluride detectors,” Lawrence Livermore National Laboratory, CA (United States), Tech. Rep. UCRL-ID-125271, 1996

    1996

  21. [29]

    A peak shape function building and comparison of cadmium zinc telluride detector for gamma-ray spectrum,

    Z. Li, R. Shi, G. Yang, J. Yang, X. Hong, and H. Zheng, “A peak shape function building and comparison of cadmium zinc telluride detector for gamma-ray spectrum,”The European Physical Journal Plus, vol. 137, no. 12, p. 1297, 2022

    2022

  22. [30]

    Charge sharing in common-grid pixelated CdZnTe detectors,

    J. C. Kim, S. E. Anderson, W. Kaye, F. Zhang, Y . Zhu, S. J. Kaye, and Z. He, “Charge sharing in common-grid pixelated CdZnTe detectors,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 654, n...

    2011

  23. [31]

    Signal modeling of charge sharing effect in simple pixelated CdZnTe detector,

    J. C. Kim, W. R. Kaye, and Z. He, “Signal modeling of charge sharing effect in simple pixelated CdZnTe detector,”Journal of the Korean Physical Society, vol. 64, no. 9, pp. 1336–1345, 2014

    2014

  24. [32]

    PointNet: Deep learning on point sets for 3D classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “PointNet: Deep learning on point sets for 3D classification and segmentation,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017

    2017

  25. [33]

    PointNet++: Deep hierarchical feature learning on point sets in a metric space,

    C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep hierarchical feature learning on point sets in a metric space,” inAdvances in Neural Information Processing Systems, I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, Eds., v...

    2017

  26. [34]

    On the representation power of set pooling networks,

    C. Bueno and A. Hylton, “On the representation power of set pooling networks,” inAdvances in Neural Information Processing Systems, M. Ranzato, A. Beygelzimer, Y . Dauphin, P. Liang, and J. W. Vaughan, Eds., vol. 34. Curran Associates, Inc., 2021, pp. 17 170– 17 182. [Online]....

    2021

  27. [35]

    How powerful are graph neural networks?

    K. Xu, W. Hu, J. Leskovec, and S. Jegelka, “How powerful are graph neural networks?” 2019. [Online]. Available: https: //arxiv.org/abs/1810.00826

    2019 arXiv

  28. [36]

    eqnn-jax,

    J. Balla, S. Mishra-Sharma, and C. Cuesta-Lazaro, “eqnn-jax,” https: //github.com/smsharma/eqnn-jax, 2024

    2024

  29. [37]

    JAX: composable transformations of Python+NumPy programs,

    J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. VanderPlas, S. Wanderman-Milne, and Q. Zhang, “JAX: composable transformations of Python+NumPy programs,” 2018. [Online]. Available: http://github.com/jax-ml/jax

    2018

  30. [38]

    Compiling machine learning programs via high-level tracing,

    R. Frostig, M. Johnson, and C. Leary, “Compiling machine learning programs via high-level tracing,” inSystems for Machine Learning (SysML) 2018, February 15–16, 2018, Stanford, CA, 2018. [Online]. Available: https://mlsys.org/Conferences/doc/2018/146.pdf

    2018

  31. [39]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,”

  32. [40]

    Axiomatic attribution for deep networks,

    M. Sundararajan, A. Taly, and Q. Yan, “Axiomatic attribution for deep networks,” inInternational conference on machine learning. PMLR, 2017, pp. 3319–3328

    2017

  33. [41]

    Data-driven performance optimization of gamma spectrometers with many channels,

    J. R. Vavrek, H. S. Parrilla, G. Aversano, M. S. Bandstra, M. Folsom, and D. Hellfeld, “Data-driven performance optimization of gamma spectrometers with many channels,”IEEE Transactions on Nuclear Science, 2026

    2026

  34. [42]

    Entry A002866 in the on-line encyclopedia of integer sequences,

    O. F. Inc., “Entry A002866 in the on-line encyclopedia of integer sequences,” 2024, retrieved from https://oeis.org/A002866 on August 6, 2024

    2024

  35. [43]

    Entry A000670 in the on-line encyclopedia of integer sequences,

    ——, “Entry A000670 in the on-line encyclopedia of integer sequences,” 2024, retrieved from https://oeis.org/A000670 on July 5, 2024

    2024

  36. [44]

    Entry A000262 in the on-line encyclopedia of integer sequences,

    ——, “Entry A000262 in the on-line encyclopedia of integer sequences,” 2024, retrieved from https://oeis.org/A000262 on July 5, 2024

    2024

  37. [45]

    Entry A000110 in the on-line encyclopedia of integer sequences,

    ——, “Entry A000110 in the on-line encyclopedia of integer sequences,” 2024, retrieved from https://oeis.org/A000110 on August 6, 2024

    2024

  38. [46]

    Dropout as a Bayesian approximation: Representing model uncertainty in deep learning,

    Y . Gal and Z. Ghahramani, “Dropout as a Bayesian approximation: Representing model uncertainty in deep learning,” inInternational conference on machine learning. PMLR, 2016, pp. 1050–1059

    2016

  39. [47]

    Evaluating and calibrating uncertainty prediction in regression tasks,

    D. Levi, L. Gispan, N. Giladi, and E. Fetaya, “Evaluating and calibrating uncertainty prediction in regression tasks,”Sensors, vol. 22, no. 15, p. 5540, 2022

    2022

  40. [48]

    Accurate uncertainties for deep learning using calibrated regression,

    V . Kuleshov, N. Fenner, and S. Ermon, “Accurate uncertainties for deep learning using calibrated regression,” inInternational conference on machine learning. PMLR, 2018, pp. 2796–2804

    2018

  41. [49]

    Uncertainty quantification in scientific machine learning: Methods, 14 metrics, and comparisons,

    A. F. Psaros, X. Meng, Z. Zou, L. Guo, and G. E. Karniadakis, “Uncertainty quantification in scientific machine learning: Methods, 14 metrics, and comparisons,”Journal of Computational Physics, vol. 477, p. 111902, 2023

    2023

  42. [50]

    Simple and scalable predictive uncertainty estimation using deep ensembles,

    B. Lakshminarayanan, A. Pritzel, and C. Blundell, “Simple and scalable predictive uncertainty estimation using deep ensembles,”Advances in neural information processing systems, vol. 30, 2017

    2017

  43. [2017]

    Available: https://arxiv.org/abs/1412.6980

    [Online]. Available: https://arxiv.org/abs/1412.6980

    arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.