Pith. sign in

REVIEW 3 major objections 2 minor 30 references

Sustainability or Survivability? Eliminating the Need to Choose in LEO Satellite Constellations

T0 review · 3 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper claims that sun-synchronous orbital planes aligned with Earth's day-night cycle can shrink LEO constellations by up to tenfold in satellite count and cut radiation exposure by roughly 23% compared with Walker-delta designs.

desk verdict The submitted PDF is two different papers: the abstract claims a sun-synchronous LEO constellation result, the body is an unrelated calibration-metrics preprint, and the headline numbers have no derivation anywhere. read the letter →

arxiv 2508.17763 v1 pith:BZMQIJJK submitted 2025-08-25 cs.NI

Chris Misa , Ramakrishnan Durairajan This is my paper · ORCID
classification cs.NI
keywords LEOsatellitenetworkssun-synchronousorbitSS-planeconstellationsWalker-deltaconstellationdiurnaltrafficdemandradiationexposuresustainabilitysurvivability
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

LEO megaconstellations are sized for worst-case global coverage, and the paper argues this is why they need tens of thousands of satellites. Its proposal, the SS-plane design, uses sun-synchronous orbit planes so that coverage follows the Earth's diurnal cycle: capacity sits over the sunlit, high-demand side of the planet and shifts around as the Earth turns. The abstract claims this reduces the required satellite count by up to an order of magnitude and cuts radiation exposure by about 23% relative to traditional Walker-delta constellations, which would mean the sustainability-versus-survivability dilemma is largely self-imposed. If the claim is right, a smaller, longer-lived, demand-matched constellation could deliver comparable global connectivity—a direct challenge to the megaconstellation build-out. A reader should know, however, that the manuscript body as provided is an unrelated preprint on calibration metrics for regression models; the constellation analysis the abstract promises is not present in the text, so the claim rests on the abstract alone.

What carries the argument

The carrying mechanism is the sun-synchronous (SS) orbit plane—an orbit whose plane keeps a fixed orientation to the Sun, so a satellite crosses each latitude at the same local solar time every day. The paper's SS-plane design assembles a constellation from such planes so that global coverage sweeps the Earth in step with the diurnal cycle: instead of blanketing the whole planet continuously, satellites concentrate over the busy, sunlit hemisphere and hand off capacity as demand rotates with the day. The paper credits this one geometric choice for both gains: matching supply to diurnal demand, which shrinks the fleet, and keeping satellites in a geometry that accumulates less radiation dose.

What would settle it

Opening the manuscript's full text settles the first question: it contains no demand model, orbit propagation, coverage analysis, or radiation-dose calculation for SS-plane constellations, because the body is a preprint on calibration metrics for regression models. To settle the substantive claim, a reader could simulate a sun-synchronous plane constellation sized against a diurnal traffic-demand map and compare the needed satellite count and accumulated radiation dose against a Walker-delta constellation of equal service quality; the claimed tenfold satellite reduction and ~23% radiation cut

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the sustainability/survivability trade-off in LEO satellite networks is an artifact of ignoring structure. Internet traffic demand has a strong spatiotemporal pattern tied to the local solar day, and the near-Earth space environment is physically uneven; a constellation built from sun-synchronous (SS) orbit planes—whose coverage is locked to the Sun so each satellite passes a given latitude at the same local time—can align capacity with that demand as the planet rotates. The claimed payoff is concrete: up to an order of magnitude fewer satellites and roughly 23% less radiation exposure relative to a Walker-delta constellation providing

Load-bearing premise

The manuscript's central claim stands on the premise that the text after the abstract actually contains the SS-plane constellation analysis, which it does not—the body is an unrelated preprint on regression-calibration metrics—and, more substantively, on the premise that global Internet demand peaks with the local solar day strongly enough for sun-synchronous planes to match it without violating coverage, latency, or revisit constraints.

Editorial extensions

If this is right

  • Constellation sizing could shift from worst-case global coverage to demand matching, with fleet sizes falling by up to an order of magnitude if the diurnal-demand claim holds.
  • A roughly 23% cut in radiation dose would extend satellite lifetimes, reducing replacement launches and space debris—directly addressing the sustainability side of the trade.
  • The survivability objection to lean constellations weakens: a demand-matched SS-plane fleet is simultaneously smaller and less radiation-exposed than the Walker-delta standard, dissolving the either/or framing.
  • The two headline numbers are coupled predictions of one geometry: the satellite-count reduction and the radiation reduction must both materialize in the same design, so either can be checked independently.
  • Megaconstellation economics would be inverted—the cheapest and most survivable network could be the one with the fewest satellites, not the most.

Reading between the lines

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

  • The claimed order-of-magnitude gain is only as strong as the diurnal structure of real traffic; inferring the paper's logic, a demand map with flat or caching-smoothed daily variation would erode the advantage, making the demand model the real object to test.
  • If SS-planes also reduce eclipse and thermal cycling, the ~23% radiation figure would likely compound with lifetime gains the paper does not state.
  • A direct test of the thesis is computable today: size two constellations against identical service targets using a measured global demand map, one in SS-planes and one Walker-delta, and compare satellite count and accumulated radiation dose.
  • The paper's framing implies the dichotomy dissolves only if coverage may 'chase' demand; regions with round-the-clock usage peaks or globally distributed traffic would be the natural failure case for the design.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 2 minor

Summary. The submitted manuscript, titled 'Sustainability or Survivability? Eliminating the Need to Choose in LEO Satellite Constellations', is presented as a networking paper proposing sun-synchronous (SS) orbit planes for LEO satellite constellations. The abstract claims that SS-plane constellations can reduce the number of required satellites by up to an order of magnitude and cut radiation exposure by ~23% relative to Walker-delta constellations, based on aligning coverage with the diurnal structure of Internet traffic. However, the full text supplied is not a satellite-constellation paper at all: it is a preprint by Wibbeke et al. on evaluating quantified uncertainty and regression calibration metrics (arXiv:2508.17761v3). The body contains no orbit mechanics, no demand model, no coverage or latency analysis, no radiation calculation, and no constellation simulation. The central quantitative claims of the abstract therefore have no supporting methodology, data, or derivation anywhere in the submitted document.

Significance. If the abstract's claims were substantiated, the contribution would be significant: an order-of-magnitude reduction in required satellite count while cutting radiation exposure would be a major result for LEO constellation sustainability and survivability, with clear implications for constellation sizing, launch costs, and orbital debris. However, the submitted manuscript provides none of the evidence needed to assess these claims. The attached body is a legitimate calibration-metrics benchmark with some strengths—controlled synthetic and real-world experiments, public benchmark code, and a systematic comparison of thirteen metrics—but it is unrelated to the satellite topic and cannot support the abstract's assertions. As it stands, the paper cannot be evaluated as a networking contribution.

major comments (3)
  1. [Abstract / Full Text] The central claims—'up to an order of magnitude' fewer satellites and '~23%' less radiation—are unsupported by the submitted body. The full text is a different manuscript, 'Evaluating the Quality of the Quantified Uncertainty for (Re)calibration of Data-Driven Regression Models' (arXiv:2508.17761v3). Sections 3–7 define and benchmark calibration metrics with equations numbered (1)–(28); none of these model satellite coverage, traffic demand, orbital mechanics, or radiation dose. There is no method or data in the document that could produce the abstract's numbers.
  2. [Abstract, second sentence] The premise that Internet traffic demand has strong, exploitable spatiotemporal structure tied to the local solar day is asserted but never modeled. No demand model, traffic dataset, or analysis is provided to show that SS-plane constellations can match demand while satisfying coverage, latency, or revisit constraints. Without such a model, the claimed reduction in satellite count is not derived.
  3. [Abstract, fourth sentence] The claimed ~23% reduction in radiation exposure has no supporting calculation. The manuscript specifies no orbital altitudes, inclinations, radiation environment model (e.g., AP/AE, SPENVIS), shielding assumptions, or mission duration. A quantitative radiation comparison between SS-plane and Walker-delta constellations cannot be reproduced or checked from the submitted text.
minor comments (2)
  1. [Throughout body] The body text contains several rendering artifacts, including non-rendered placeholder characters near Figures 2 and 6–8 and an omitted GitHub URL for the benchmark code. These issues hinder readability, although they are secondary to the core problem of the missing satellite content.
  2. [Title and metadata] The arXiv metadata, title, abstract, and body are inconsistent: the title and abstract describe LEO satellite networks, while the body and author list describe a calibration-metrics preprint. The submission must be corrected to match the intended manuscript.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the claimed SS-plane derivation is absent from the supplied text, so there is no derivation chain that reduces to its own inputs.

full rationale

The submitted manuscript consists of two disjoint components. The abstract and front matter claim an SS-plane LEO satellite constellation analysis (arXiv:2508.17763), stating: "We demonstrate that SS-plane constellations can reduce the number of satellites required by up to an order of magnitude and cut radiation exposure by ~23% compared to traditional Walker-delta constellations." The full text, however, is an unrelated calibration-metrics preprint by Wibbeke et al. (arXiv:2508.17761v3), which begins "EVALUATING THE QUALITY OF THE QUANTIFIED UNCERTAINTY FOR (RE)CALIBRATION OF DATA-DRIVEN REGRESSION MODELS" and contains no orbit model, demand model, coverage analysis, radiation-dose calculation, or constellation simulation. Because the satellite claim has no derivation in the body, there is no equation, fitted parameter, or definitional chain that can be shown to be equivalent to an input by construction. Unsupported claims and document mismatches are evidentiary or correctness problems, not circularity. Within the actual body, the recommendation that ENCE and CWC are the most dependable metrics is supported by a benchmark with public code, real-world and synthetic datasets, and controlled miscalibration experiments; it is not generated by self-citation or by defining the metrics in terms of the conclusion. The few self-citations (e.g., Wibbeke et al. 2025) are incidental and not load-bearing. Therefore no significant circularity is present, and the appropriate score is 0.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

This ledger is dominated by the document mismatch. The abstract's two headline numbers are free claims: no derivation, fitting procedure, or data source is given for the 10x satellite reduction or the 23% radiation reduction. Parameters that do have a home live in the unrelated body (perturbation scales, the 3% detection threshold, eta=50, 10 bins), which affects only the calibration benchmark and not the stated satellite claim. The axioms listed are likewise split: two belong to the satellite abstract's unstated demand model, and three belong to the body's benchmark protocol. A reader pays for none of this upstream.

free parameters (6)
  • Claimed satellite reduction factor = up to ~10x
    Abstract states SS-plane constellations reduce required satellite count by up to an order of magnitude; no method or simulation is provided, so the figure is a free claim.
  • Claimed radiation exposure reduction = ~23%
    Abstract states ~23% radiation reduction; no dosimetry model or orbit simulation is provided.
  • Miscalibration perturbation scale (body) = 0.9 to 1.1
    Section 4.4 and Eq. 25-28: authors choose scaling factors 0.9 and 1.1 to define artificial miscalibration; results depend on these choices.
  • Detection threshold (body) = 3% relative change
    Section 4.4: a miscalibration is deemed correctly identified only if the metric changes by at least 3%; this threshold shapes the headline ENCE/CWC recommendation.
  • CWC penalty parameter (body) = eta = 50
    Section 3.2 and Appendix A: adopted from Khosravi et al. 2011; CWC values and hence the CWC ranking depend on it.
  • Bin count for binned metrics (body) = 10 bins
    Appendix A: ENCE, UCE, QCE, ECPE use 10 bins; authors report robustness checks at 7 and 15 bins in Section 5.
assumptions (4)
  • domain assumption Gaussian predictive distributions for NLL, CRPS and ENCE-based metrics
    Sections 3.4-3.6 assume predicted uncertainty is Gaussian to derive closed forms; the body acknowledges this in Section 5.
  • domain assumption Internet traffic demand is strongly periodic with the local solar day and can be served by sun-synchronous coverage
    Abstract, second sentence: the entire SS-plane rationale rests on this premise; no demand model is presented.
  • ad hoc to paper Perfect calibration of the synthetic miscalibration generator
    Section 4.4, Eq. 25-28: the sinusoidal heteroscedastic noise model is asserted to approximate realistic conditions and to be perfectly calibrated.
  • domain assumption Deep ensemble with 5 MLPs is a sufficient probe of metric behavior
    Section 4: all benchmarks use one architecture; the body argues metrics are model-agnostic, which is asserted, not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sustainability or Survivability? Eliminating the Need to Choose in LEO Satellite Constellations." pith.science (2026). https://pith.science/paper/BZMQIJJK

@misc{pith2026250817763,
  author       = {Pith},
  title        = {Pith review of: Sustainability or Survivability? Eliminating the Need to Choose in LEO Satellite Constellations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BZMQIJJK}},
  note         = {Machine review of arXiv:2508.17763}
}
read the original abstract

LEO Satellite Networks (LSNs) are revolutionizing global connectivity, but their reliance on tens of thousands of satellites raises pressing concerns over sustainability and survivability. In this work, we argue that the inefficiencies in LSN designs stem from ignoring the strong spatiotemporal structure of Internet traffic demand (which impacts sustainability) and the physical realities of the near-Earth space environment (which affects survivability). We propose a novel design approach based on sun-synchronous (SS) orbits called SS-plane, which aligns satellite coverage with the Earth's diurnal cycle. We demonstrate that SS-plane constellations can reduce the number of satellites required by up to an order of magnitude and cut radiation exposure by ~23% compared to traditional Walker-delta constellations. These findings suggest a paradigm shift in LSN research from large, disposable megaconstellations to more sustainable, targeted LEO constellations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 9 canonical work pages

  1. [1]

    4������������������������������������������������������� 5�������������������������������������� 6���������������������������� 25

    To avoid model distortions due to exceeding noise or erroneous samples, all datasets are cleaned of outliers using an isolation forest outlier detection with a probability threshold of 0.8 [Liu et al., 2012]. 4������������������������������������������������������� 5�������������������������������������� 6���������������������������� 25

    work page 2012

  2. [3]

    Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U

    doi:10.1109/MCI.2022.3155327. Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U. Rajendra Acharya, Vladimir Makarenkov, and Saeid Nahavandi. A review of uncertainty quantification in deep learning: Techniques, applications and challenges.Information Fusion, 76: 243–...

    work page arXiv 2022

  3. [7]

    Laurens Sluijterman, Eric Cator, and Tom Heskes

    doi:10.1007/s10994-021-05946-3. Laurens Sluijterman, Eric Cator, and Tom Heskes. How to evaluate uncertainty estimates in machine learning for regression?Neural Networks, 173:106203,

    doi:10.1007/s10994-021-05946-3

  4. [9]

    Apostolos F

    arXiv:2302.13425 [cs]. Apostolos F. Psaros, Xuhui Meng, Zongren Zou, Ling Guo, and George Em Karniadakis. Uncertainty quantification in scientific machine learning: Methods, metrics, and comparisons.Journal of Computational Physics, 477:111902, March

    arXiv

  5. [10]

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q

    doi:10.1016/j.jcp.2022.111902. Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. On Calibration of Modern Neural Networks. In Proceedings of the 34th International Conference on Machine Learning, volume 70 ofProceedings of Machine Learning Research, pages 1321–1330. PMLR, August

    work page arXiv 2022

  6. [11]

    Victor Dheur and Souhaib Ben Taieb

    doi:10.3390/s22155540. Victor Dheur and Souhaib Ben Taieb. A Large-Scale Study of Probabilistic Calibration in Neural Network Regression. InProceedings of the 40th International Conference on Machine Learning, volume 202 ofProceedings of Machine Learning Research, pages 7813–7836. PMLR, July

    doi:10.3390/s22155540

  7. [12]

    Parametric and Multivariate Uncertainty Calibration for Regression and Object Detection

    Fabian Küppers, Jonas Schneider, and Anselm Haselhoff. Parametric and Multivariate Uncertainty Calibration for Regression and Object Detection. InComputer Vision – ECCV 2022 Workshops, pages 426–442, Cham,

    2022

  8. [13]

    doi:10.1371/journal.pcbi.1011392. A. Khosravi, S. Nahavandi, D. Creighton, and A. F. Atiya. Comprehensive Review of Neural Network-Based Prediction Intervals and New Advances.IEEE Transactions on Neural Networks, 22(9):1341–1356, September

    doi:10.1371/journal.pcbi.1011392

Show all 30 references
  1. [15]

    doi:10.1016/j.patrec.2024.09.011

    ISSN 01678655. doi:10.1016/j.patrec.2024.09.011. Tilmann Gneiting and Adrian E Raftery. Strictly Proper Scoring Rules, Prediction, and Estimation.Journal of the American Statistical Association, 102(477):359–378, March

    2024 doi

  2. [17]

    Shengjia Zhao, Tengyu Ma, and Stefano Ermon

    arXiv:2005.12496 [cs]. Shengjia Zhao, Tengyu Ma, and Stefano Ermon. Individual Calibration with Randomized Forecasting. InProceedings of the 37th International Conference on Machine Learning, volume 119 ofProceedings of Machine Learning Research, pages 11387–11397. PMLR, July

    2005 arXiv

  3. [18]

    Youngseog Chung, Ian Char, Han Guo, Jeff Schneider, and Willie Neiswanger

    arXiv:2506.01486 [cs.LG]. Youngseog Chung, Ian Char, Han Guo, Jeff Schneider, and Willie Neiswanger. Uncertainty Toolbox: an Open-Source Library for Assessing, Visualizing, and Improving Uncertainty Quantification, 2021b. arXiv:2109.10254 [cs.LG]. F. Pedregosa, G. Varoquaux, A...

    arXiv

  4. [23]

    Karl Ulrich

    https://doi.org/10.24432/C5PG66. Karl Ulrich. Servo [dataset]. UCI Machine Learning Repository, 1993.https://doi.org/10.24432/C5Q30F. I-Cheng Yeh. Concrete compressive strength [dataset]. UCI Machine Learning Repository,

    1993 doi

  5. [24]

    Sathishkumar V .E., Changsun Shin, and Yongyun Cho

    https://doi.org/ 10.24432/C5PK67. Sathishkumar V .E., Changsun Shin, and Yongyun Cho. Efficient energy consumption prediction model for a data analytic-enabled industry building in a smart city [dataset].Building Research & Information, 49(1):127–143,

    doi

  6. [25]

    Athanasios Tsanas, Max A

    https://doi.org/10.24432/C5002N. Athanasios Tsanas, Max A. Little, Patrick E. McSharry, and Lorraine O. Ramig. Accurate telemonitoring of parkinson’s disease progression by noninvasive speech tests [dataset].IEEE Transactions on Biomedical Engineering, 57(4): 884–893,

    doi

  7. [29]

    All synthetic datasets consist of 1000 samples and are generated without noise

    Some of the synthetic datasets are created using the sklearn library [Pedregosa et al., 2011]. All synthetic datasets consist of 1000 samples and are generated without noise. For both dataset types, all features are normalized by minmax-scaling between 0 and

    2011

  8. [1987]

    24432/C5830D

    https://doi.org/10. 24432/C5830D. Warwick Nash, Tracy Sellers, Simon Talbot, Andrew Cawthorn, and Wes Ford. Abalone [dataset]. UCI Machine Learning Repository, 1995.https://doi.org/10.24432/C55C7W. Paulo Cortez, António Cerdeira, Fernando Almeida, Telmo Matos, and José Reis. M...

    1995 doi

  9. [2007]

    Eric Zelikman, Christopher Healy, Sharon Zhou, and Anand Avati

    doi:10.1198/016214506000001437. Eric Zelikman, Christopher Healy, Sharon Zhou, and Anand Avati. CRUDE: Calibrating Regression Uncertainty Distributions Empirically, March

    doi

  10. [2008]

    Srgio Moro, Paulo Rita, and Bernardo Vala

    https://doi.org/ 10.24432/C5D88D. Srgio Moro, Paulo Rita, and Bernardo Vala. Facebook metrics [dataset]. UCI Machine Learning Repository,

    doi

  11. [2010]

    Elaheh Ordoni, Jakob Bach, and Ann-Katrin Fleck

    doi:10.1109/TBME.2009.2036000. Elaheh Ordoni, Jakob Bach, and Ann-Katrin Fleck. Analyzing and predicting verification of data-aware process models–a case study with spectrum auctions [dataset].IEEE Access, 10:31699–31713,

    2009

  12. [2011]

    23 arXivTemplateA PREPRINT Simon Kristoffersson Lind, Ziliang Xiong, Per-Erik Forssén, and V olker Krüger

    doi:10.1109/TNN.2011.2162110. 23 arXivTemplateA PREPRINT Simon Kristoffersson Lind, Ziliang Xiong, Per-Erik Forssén, and V olker Krüger. Uncertainty quantification metrics for deep regression.Pattern Recognition Letters, 186:91–97, October

    2011

  13. [2012]

    An Dinh, Stacey Miertschin, Amber Young, and Somya D

    https://doi.org/10.24432/C51307. An Dinh, Stacey Miertschin, Amber Young, and Somya D. Mohanty. A data-driven approach to predicting diabetes and cardiovascular disease with machine learning [dataset].BMC medical informatics and decision making, 19(1):1–15,

    doi

  14. [2014]

    Kam Hamidieh

    https://doi.org/10.24432/C5VW2C. Kam Hamidieh. Superconductivty data [dataset]. UCI Machine Learning Repository,

    doi

  15. [2016]

    Jacob Feldmesser

    https://doi.org/10.24432/C5QK55. Jacob Feldmesser. Computer hardware [dataset]. UCI Machine Learning Repository,

    doi

  16. [2018]

    doi:10.1109/ACCESS.2018.2836917. Jakob Gawlikowski, Cedrique Rovile Njieutcheu Tassi, Mohsin Ali, Jongseok Lee, Matthias Humt, Jianxiang Feng, Anna Kruspe, Rudolph Triebel, Peter Jung, Ribana Roscher, Muhammad Shahzad, Wen Yang, Richard Bamler, and Xiao Xiang Zhu. A survey of ...

    2018

  17. [2019]

    net:cal - Uncertainty Calibration

    24 arXivTemplateA PREPRINT Table 2: List of used real-world datasets. No. (ID) Name Number of samples after cleaning Ref. 1 forest-fires 513 [Cortez and Morais, 2008] 2 facebook-metrics 493 [Moro et al., 2016] 3 computer-hardware 206 [Feldmesser, 1987] 4 abalone 4145 [Nash et ...

    2008

  18. [2021]

    doi:10.1016/j.inffus.2021.05.008. H. M. Dipu Kabir, Abbas Khosravi, Mohammad Anwar Hosen, and Saeid Nahavandi. Neural Network-Based Uncertainty Quantification: A Survey of Methodologies and Applications.IEEE Access, 6:36218–36234,

    2021 doi

  19. [2022]

    doi:10.1016/j.eswa.2022.116659

    ISSN 0957-4174. doi:10.1016/j.eswa.2022.116659. Yarin Gal and Zoubin Ghahramani. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. InProceedings of The 33rd International Conference on Machine Learning, volume 48 ofProceedings of Machine Lea...

    2022

  20. [2023]

    doi:10.1007/s10462-023-10562-9

    ISSN 1573-7462. doi:10.1007/s10462-023-10562-9. Eyke Hüllermeier and Willem Waegeman. Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods.Machine Learning, 110(3):457–506, March

    doi

  21. [2024]

    Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, and Yukun Li

    doi:10.1016/j.neunet.2024.106203. Wenchong He, Zhe Jiang, Tingsong Xiao, Zelin Xu, and Yukun Li. A Survey on Uncertainty Quantification Methods for Deep Learning, January

    2024

  22. [2025]

    Max-Heinrich Laves, Sontje Ihler, Jacob F

    doi:10.1109/TITS.2025.3532803. Max-Heinrich Laves, Sontje Ihler, Jacob F. Fast, Lüder A. Kahrs, and Tobias Ortmaier. Well-Calibrated Regression Uncertainty in Medical Imaging with Deep Learning. InProceedings of the Third Conference on Medical Imaging with Deep Learning, volum...

    2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.