Pith. sign in

REVIEW 3 major objections 1 minor 1 cited by

Multidimensional Distributional Neural Network Output Demonstrated in Super-Resolution of Surface Wind Speed

T0 review · 3 major / 1 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 This paper presents a framework for training neural networks with a multidimensional Gaussian loss, letting them output closed-form predictive distributions with explicit spatial correlation over super-resolved wind speed fields.

desk verdict Plausible, useful-sounding framework for correlated predictive uncertainty, but the supplied full text is a different paper and the abstract alone gives no numbers, baselines, or calibration evidence to judge soundness. read the letter →

arxiv 2508.16686 v1 pith:7WSHDGRP submitted 2025-08-21 cs.LG stat.MEstat.ML

Harrison J. Goldwyn , Mitchell Krock , Johann Rudi , Daniel Getter , Julie Bessac This is my paper
classification cs.LGstat.MEstat.ML
keywords uncertaintyquantificationmultidimensionalGaussianlosssuper-resolutionsurfacewindspeedcovarianceestimationFourierrepresentationdistributionalneuralnetworkheteroscedasticity
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 tries to establish that a regression network can produce not just a point prediction for each output pixel, but a full multivariate Gaussian distribution over the entire high-dimensional output, complete with a learned covariance that encodes how errors at different locations move together. It shows this is possible on a surface wind speed super-resolution task by representing the covariance in the Fourier domain, which keeps training stable and preserves spatial correlation, and by adding an information-sharing regularizer that interpolates between image-specific and global covariance estimates. The stated payoff is tractable, explicitly correlated aleatoric uncertainty for scientific prediction, without sacrificing ordinary prediction performance, and a scheme that extends to non-Gaussian distribution families.

What carries the argument

The load-bearing objects are the multidimensional Gaussian loss and the covariance parameterization. The mean and covariance are iteratively estimated during training, the covariance is represented in the Fourier domain to keep the matrix tractable and to enforce near-translation-invariant spatial correlation, and the 'information sharing' regularizer blends each image's covariance with a global covariance estimate to prevent the image-specific distributional loss from failing to converge. Together these pieces let the network emit a closed-form multivariate Gaussian predictive distribution with learned spatial error correlation.

What would settle it

On a held-out set of observed wind-speed fields, compute the empirical covariance of prediction errors and compare it with the model's predictive covariance, for example via a quadratic score or a multivariate calibration metric; if the model's off-diagonal correlations are not closer to the empirical error correlations than a diagonal covariance would be, the central claim that spatial correlation is preserved fails.

Watch

Extended reading notes

Core claim

The central claim is that a neural network can be trained with a multidimensional Gaussian negative log-likelihood whose mean and covariance are estimated iteratively, yielding a closed-form predictive distribution with heteroscedastic, non-identically distributed structure across output locations. On the wind-speed downscaling example, the covariance is parameterized in the Fourier domain to stabilize training and preserve spatial correlation, while an information-sharing regularizer interpolates between image-specific and global covariance estimates—this interpolation is what allows the image-specific distributional loss to converge at all. The paper asserts that this framework supports ef

Load-bearing premise

The load-bearing premise is that the information-sharing regularizer—interpolating between image-specific and global covariance estimates—produces covariances that reflect each image's true error-correlation structure, rather than merely being stabilized by shrinking toward a global average.

Editorial extensions

If this is right

  • Super-resolved wind speed fields come with a covariance matrix that describes how prediction errors at different grid cells are correlated, not just individual pixel variances.
  • Downstream users can draw full-field samples from the predictive distribution, yielding many consistent wind maps rather than independent per-pixel perturbations.
  • The Fourier covariance representation keeps the method computationally feasible for high-dimensional outputs and stabilizes training of the distributional loss.
  • The information-sharing regularizer provides a route to train image-specific per-scene uncertainty models that would otherwise diverge.
  • The same loss and covariance machinery should transfer to other high-dimensional scientific prediction tasks and to more complex, non-Gaussian predictive distribution families.

Reading between the lines

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

  • The Fourier covariance representation implies a near-stationary, translation-invariant error correlation structure; if wind errors are genuinely location-dependent (near terrain or coastlines), the predictive uncertainty will smooth that variation away—a testable limitation.
  • Because the information-sharing weight interpolates toward a global covariance, the reported uncertainty inherits a dependence on that hyperparameter; out-of-sample calibration of the covariance, rather than only the marginal variances, is a natural check the abstract does not yet report.
  • The method as described captures aleatoric uncertainty only; combining it with an ensemble of networks would be a direct extension to also represent epistemic uncertainty.
  • The framework's claim of 'without disrupting prediction performance' suggests the covariance loss could be added to existing deterministic super-resolution models as a plug-in head, which would make the correlated uncertainty broadly available.
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 / 1 minor

Summary. The abstract of arXiv:2508.16686 proposes a framework for training neural networks with a multidimensional Gaussian loss, yielding closed-form predictive distributions with spatial correlation for super-resolution of surface wind speed. It introduces a Fourier representation of the covariance matrix and an 'information sharing' regularizer that interpolates between image-specific and global covariance estimates, claiming stable training, efficient sampling, and preservation of prediction performance. However, the supplied full text is an unrelated condensed-matter physics paper on doped triangular lattice Kitaev magnets, containing no equations, experiments, baselines, or results that support the abstract's claims. Thus, the manuscript as provided cannot be evaluated as a machine-learning paper.

Significance. If the abstract's claims are correct, the framework could contribute meaningfully to uncertainty quantification in scientific regression tasks, particularly for outputs with correlated spatial structure. The idea of a closed-form learned multivariate Gaussian covariance with a tractable Fourier representation and an information-sharing stabilizer is potentially useful. However, the significance cannot be assessed from the submitted materials: there is no technical content in the full text that corresponds to the abstract, no empirical evaluation, no calibration analysis, and no comparison to existing methods. The paper currently provides no verifiable evidence for its central claims.

major comments (3)
  1. [Full text] The full text supplied is a physics paper titled 'Searching for superconductivity in doped triangular lattice Kitaev magnets' and does not contain any of the methodology, equations, experiments, or results described in the abstract. No derivation of the multidimensional Gaussian loss, Fourier covariance representation, information-sharing regularizer, or super-resolution experiments is present. This is a load-bearing absence: the central claims of the abstract are entirely unsupported by the submitted manuscript.
  2. [Abstract] The abstract claims that the framework 'stabilizes training,' 'preserves spatial correlation,' and does so 'without disrupting prediction performance,' but no numerical results, baseline comparisons, or quantitative statements are provided anywhere in the supplied text. Even if the full text were the intended paper, such claims would require empirical support; as submitted, there is no evidence to evaluate.
  3. [Abstract (information sharing)] The abstract concedes that image-specific distributional loss functions without the information-sharing regularizer fail to converge. The regularizer's mixing weight is a tuning parameter whose effect on the fidelity of the resulting covariance is not analyzed, and the Fourier representation imposes near-stationary structure on spatial correlations. No out-of-sample calibration or comparison against empirical error correlations is reported, so it is unclear whether the uncertainty estimates are faithful or merely a training-stabilization artifact.
minor comments (1)
  1. [Abstract] The phrase 'non-identically distributed and heteroscedastic structure' is redundant; 'non-identically distributed' already implies heteroscedasticity in the marginal sense. The abstract also uses 'without disrupting prediction performance' without defining what disruption means or how prediction performance is measured.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the abstract describes standard likelihood-based covariance estimation.

full rationale

The abstract's central claim is that a neural network trained with a multidimensional Gaussian loss yields closed-form predictive distributions with learned covariance matrices. The covariance is a likelihood parameter estimated from training data; no equation in the abstract defines the prediction as a function of the same fitted quantity. The 'information sharing' regularizer interpolates between image-specific and global covariance estimates to stabilize training; this is a standard regularization technique and not a fitted input renamed as a prediction. The Fourier representation is a modeling assumption about covariance structure, not a circular definition. There are no self-citations or imported uniqueness theorems in the abstract. The provided full text is a different manuscript (on superconductivity in Kitaev magnets) and contains no derivation chain for the wind-speed framework; this mismatch is a serious completeness issue but not evidence of circularity. All concerns from the skeptic summary—lack of calibration evidence, unquantified interpolation weight, stationarity assumption—are validity/robustness concerns, not circularity. For a circularity finding, a specific reduction (e.g., Eq. X = Eq. Y by construction) must be exhibited; none is present.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

For an abstract-only review, the ledger lists what the framework imports without proof from the abstract alone: a Gaussian ansatz for predictions, a stationarity assumption hidden in the Fourier covariance, and a convergence guarantee that depends on the information-sharing regularizer. Two tuning knobs, the interpolation weight and the spectral truncation, are free parameters not quantified in the abstract. No invented entities are posted.

free parameters (2)
  • information_sharing_mixing_weight
    The abstract introduces information sharing as an interpolation between image-specific and global covariance estimates; the interpolation coefficient is not given and is presumably tuned to make training converge.
  • Fourier_covariance_truncation_level
    Using a Fourier representation of the covariance matrix implies a choice of how many spectral modes are kept; this truncation controls the trade-off between computational tractability and fidelity of spatial correlation. Not specified in the abstract.
assumptions (3)
  • domain assumption The conditional distribution of the target, surface wind speed super-resolution output, is adequately modeled as multivariate Gaussian.
    The framework's closed-form predictive distribution is Gaussian; the abstract provides no evidence that wind-speed residuals are Gaussian, light-tailed, or unimodal.
  • domain assumption The Fourier representation of the covariance matrix preserves the genuinely heterogeneous spatial correlations of wind fields.
    Fourier-domain covariance implies translation-invariant, stationary or circulant, structure; the abstract offers no justification that this captures spatially varying correlation in wind fields.
  • domain assumption Iterative estimation of means and covariance matrices converges during training; the abstract itself notes this fails without information sharing.
    The abstract states information sharing was introduced because image-specific distributional losses did not converge, so the convergence guarantee rests on this regularizer.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multidimensional Distributional Neural Network Output Demonstrated in Super-Resolution of Surface Wind Speed." pith.science (2026). https://pith.science/paper/7WSHDGRP

@misc{pith2026250816686,
  author       = {Pith},
  title        = {Pith review of: Multidimensional Distributional Neural Network Output Demonstrated in Super-Resolution of Surface Wind Speed},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7WSHDGRP}},
  note         = {Machine review of arXiv:2508.16686}
}
read the original abstract

Accurate quantification of uncertainty in neural network predictions remains a central challenge for scientific applications involving high-dimensional, correlated data. While existing methods capture either aleatoric or epistemic uncertainty, few offer closed-form, multidimensional distributions that preserve spatial correlation while remaining computationally tractable. In this work, we present a framework for training neural networks with a multidimensional Gaussian loss, generating closed-form predictive distributions over outputs with non-identically distributed and heteroscedastic structure. Our approach captures aleatoric uncertainty by iteratively estimating the means and covariance matrices, and is demonstrated on a super-resolution example. We leverage a Fourier representation of the covariance matrix to stabilize network training and preserve spatial correlation. We introduce a novel regularization strategy -- referred to as information sharing -- that interpolates between image-specific and global covariance estimates, enabling convergence of the super-resolution downscaling network trained on image-specific distributional loss functions. This framework allows for efficient sampling, explicit correlation modeling, and extensions to more complex distribution families all without disrupting prediction performance. We demonstrate the method on a surface wind speed downscaling task and discuss its broader applicability to uncertainty-aware prediction in scientific models.

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Neural posterior estimation for scalable and accurate inverse parameter inference in Li-ion batteries

    physics.data-an 2026-04 unverdicted novelty 5.0 of 10

    NPE delivers millisecond-scale parameter inference for Li-ion batteries that matches or exceeds Bayesian calibration accuracy while adding local sensitivity interpretability, though with higher voltage prediction errors.

Reference graph

Works this paper leans on

46 extracted references · 39 canonical work pages · cited by 1 Pith paper

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter doi edition editor eid eprint howpublished department institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 '...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION if.digit duplicate "0" = swap duplicate "1" = swap duplicate "2" = swap duplicate "3" = swap duplicate "4" = swap duplicate "5" = swap duplicate "6" = swap duplicate "7" = swap duplicate "8" = swap "9" = or or or or or or or or or FUNCTION n.separate 't := "" #0 'numnames := t empty not t #-1 #1 subs...

  3. [3]

    , Kang, B

    ahn2018fast APACrefauthors Ahn, N. , Kang, B. \ Sohn, K A. APACrefauthors \ 2018 . Fast, accurate, and lightweight super-resolution with cascading residual network Fast, accurate, and lightweight super-resolution with cascading residual network . Proceedings of the European conference on computer vision (ECCV) Proceedings of the european conference on com...

    work page 2018

  4. [4]

    Algorithmic Hallucinations of Near-Surface Winds: Statistical Downscaling with Generative Adversarial Networks to Convection-Permitting Scales

    annau+cm23 APACrefauthors Annau, N J. , Cannon, A J. \ Monahan, A H. APACrefauthors \ 2023 . Algorithmic Hallucinations of Near-Surface Winds: Statistical Downscaling with Generative Adversarial Networks to Convection-Permitting Scales Algorithmic Hallucinations of Near-Surface Winds: Statistical Downscaling with Generative Adversarial Networks to Convect...

    work page Pith review arXiv 2023

  5. [5]

    \ Titterington, D M

    cheng1994neural APACrefauthors Cheng, B. \ Titterington, D M. APACrefauthors \ 1994 . Neural networks: A review from a statistical perspective Neural networks: A review from a statistical perspective . Statistical science 2--30

    work page 1994

  6. [6]

    , Choi, U J

    conde2022swin2sr APACrefauthors Conde, M V. , Choi, U J. , Burchi, M. \ Timofte, R. APACrefauthors \ 2022 . Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration . European Conference on Computer Vision European conference on computer vision \ ( \...

    work page 2022

  7. [7]

    Capturing Climatic Variability: Using Deep Learning for Stochastic Downscaling

    daust+m24 APACrefauthors Daust, K. \ Monahan, A. APACrefauthors \ 2024 . Capturing Climatic Variability: Using Deep Learning for Stochastic Downscaling Capturing Climatic Variability: Using Deep Learning for Stochastic Downscaling . arXiv:2406.02587

    work page Pith review arXiv 2024

  8. [8]

    , Loy, C C

    dong2014learning APACrefauthors Dong, C. , Loy, C C. , He, K. \ Tang, X. APACrefauthors \ 2014 . Learning a deep convolutional network for image super-resolution Learning a deep convolutional network for image super-resolution . Computer Vision--ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part IV 13 Compute...

    work page 2014

Show all 46 references
  1. [9]

    , Pawar, N

    faroughi2022physics APACrefauthors Faroughi, S A. , Pawar, N. , Fernandes, C. , Raissi, M. , Das, S. , Kalantari, N K. \ Mahjour, S K. APACrefauthors \ 2022 . Physics-guided, physics-informed, and physics-encoded neural networks in scientific computing Physics-guided, physics-...

    2022 arXiv

  2. [10]

    , Tassi, C R N

    gawlikowski2023survey APACrefauthors Gawlikowski, J. , Tassi, C R N. , Ali, M. , Lee, J. , Humt, M. , Feng, J. others APACrefauthors \ 2023 . A survey of uncertainty in deep neural networks A survey of uncertainty in deep neural networks . Artificial Intelligence Review 56 Sup...

    2023

  3. [11]

    , Bessac, J

    getter2024statistical APACrefauthors Getter, D. , Bessac, J. , Rudi, J. \ Feng, Y. APACrefauthors \ 2024 . Statistical Treatment of Convolutional Neural Network Superresolution of Inland Surface Wind for Subgrid-Scale Variability Quantification Statistical treatment of convolu...

    2024

  4. [12]

    , Zhang, X

    he2016deep APACrefauthors He, K. , Zhang, X. , Ren, S. \ Sun, J. APACrefauthors \ 2016 . Deep residual learning for image recognition Deep residual learning for image recognition . Proceedings of the IEEE conference on computer vision and pattern recognition Proceedings of the...

    2016

  5. [13]

    , Woolnough, S J

    holloway2012 APACrefauthors Holloway, C E. , Woolnough, S J. \ Lister, G M S. APACrefauthors \ 2012 . Precipitation distributions for explicit versus parametrized convection in a large-domain high-resolution tropical case study Precipitation distributions for explicit versus p...

    2012

  6. [14]

    , Woolnough, S J

    holloway2013 APACrefauthors Holloway, C E. , Woolnough, S J. \ Lister, G M S. APACrefauthors \ 2013 . The effects of explicit versus parameterized convection on the MJO in a large-domain high-resolution tropical case study, part I : C haracterization of large-scale organizatio...

    2013

  7. [15]

    , Woolnough, S J

    holloway2015 APACrefauthors Holloway, C E. , Woolnough, S J. \ Lister, G M S. APACrefauthors \ 2015 . The effects of explicit versus parameterized convection on the MJO in a large-domain high-resolution tropical case study, part II : P rocesses leading to differences in MJO de...

    2015

  8. [16]

    , Naiel, M A

    hu2019runet APACrefauthors Hu, X. , Naiel, M A. , Wong, A. , Lamm, M. \ Fieguth, P. APACrefauthors \ 2019 . RUNet: A robust UNet architecture for image super-resolution RUNet: A robust UNet architecture for image super-resolution . Proceedings of the IEEE/CVF Conference on Com...

    2019

  9. [17]

    \ Hwang, J N

    hu2018handbook APACrefauthors Hu, Y H. \ Hwang, J N. APACrefauthors \ 2018 . Handbook of neural network signal processing Handbook of neural network signal processing . CRC press

    2018

  10. [18]

    \ Oates, C J

    karvonen2023maximum APACrefauthors Karvonen, T. \ Oates, C J. APACrefauthors \ 2023 . Maximum likelihood estimation in Gaussian process regression is ill-posed Maximum likelihood estimation in Gaussian process regression is ill-posed . Journal of Machine Learning Research 24 120 1--47

    2023

  11. [19]

    , Lee, J K

    kim2016accurate APACrefauthors Kim, J. , Lee, J K. \ Lee, K M. APACrefauthors \ 2016 . Accurate image super-resolution using very deep convolutional networks Accurate image super-resolution using very deep convolutional networks . Proceedings of the IEEE conference on computer...

    2016

  12. [20]

    APACrefauthors \ 2017

    kutz2017deep APACrefauthors Kutz, J N. APACrefauthors \ 2017 . Deep learning in fluid dynamics Deep learning in fluid dynamics . Journal of Fluid Mechanics 814 1--4

    2017

  13. [21]

    , Theis, L

    ledig2017photo APACrefauthors Ledig, C. , Theis, L. , Husz \'a r, F. , Caballero, J. , Cunningham, A. , Acosta, A. others APACrefauthors \ 2017 . Photo-realistic single image super-resolution using a generative adversarial network Photo-realistic single image super-resolution ...

    2017

  14. [22]

    , Yang, Y

    li2022srdiff APACrefauthors Li, H. , Yang, Y. , Chang, M. , Chen, S. , Feng, H. , Xu, Z. Chen, Y. APACrefauthors \ 2022 . SRDiff: Single Image Super-Resolution with Diffusion Probabilistic Models SRDiff: Single Image Super-Resolution with Diffusion Probabilistic Models . Neuro...

    2022

  15. [23]

    , Cao, J

    liang2021swinir APACrefauthors Liang, J. , Cao, J. , Sun, G. , Zhang, K. , Van Gool, L. \ Timofte, R. APACrefauthors \ 2021 . SwinIR: Image Restoration Using Swin Transformer SwinIR: Image Restoration Using Swin Transformer . Proceedings of the IEEE/CVF international conferenc...

    2021

  16. [24]

    \ Lee, K M

    lim2017deep APACrefauthors Lim, B. \ Lee, K M. APACrefauthors \ 2017 . Deep recurrent ResNet for video super-resolution Deep recurrent resnet for video super-resolution . 2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) ...

    2017

  17. [25]

    , Son, S

    lim2017enhanced APACrefauthors Lim, B. , Son, S. , Kim, H. , Nah, S. \ Mu Lee, K. APACrefauthors \ 2017 . Enhanced deep residual networks for single image super-resolution Enhanced deep residual networks for single image super-resolution . Proceedings of the IEEE conference on...

    2017

  18. [26]

    , Racah, E

    liu2016application APACrefauthors Liu, Y. , Racah, E. , Correa, J. , Khosrowshahi, A. , Lavers, D. , Kunkel, K. others APACrefauthors \ 2016 . Application of deep convolutional neural networks for detecting extreme weather in climate datasets Application of deep convolutional ...

    2016 arXiv

  19. [27]

    \ Romo, J

    lopez2009concept APACrefauthors L \'o pez-Pintado, S. \ Romo, J. APACrefauthors \ 2009 . On the concept of depth for functional data On the concept of depth for functional data . Journal of the American statistical Association 104 486 718--734

    2009

  20. [28]

    , Matthews, A J

    love2011 APACrefauthors Love, B S. , Matthews, A J. \ Lister, G M S. APACrefauthors \ 2011 . The diurnal cycle of precipitation over the Maritime Continent in a high-resolution atmospheric model The diurnal cycle of precipitation over the Maritime Continent in a high-resolutio...

    2011 doi

  21. [29]

    , Asperti, A

    merizzi2024wind APACrefauthors Merizzi, F. , Asperti, A. \ Colamonaco, S. APACrefauthors \ 2024 . Wind speed super-resolution and validation: from ERA5 to CERRA via diffusion models Wind speed super-resolution and validation: from ERA5 to CERRA via diffusion models . Neural Co...

    2024

  22. [30]

    , Riche, R L

    mohammadi2016analytic APACrefauthors Mohammadi, H. , Riche, R L. , Durrande, N. , Touboul, E. \ Bay, X. APACrefauthors \ 2016 . An analytic comparison of regularization methods for Gaussian processes An analytic comparison of regularization methods for Gaussian processes . arX...

    2016 arXiv

  23. [31]

    , Tahmassebi, A

    mohebali2020probabilistic APACrefauthors Mohebali, B. , Tahmassebi, A. , Meyer-Baese, A. \ Gandomi, A H. APACrefauthors \ 2020 . Probabilistic neural networks: a brief overview of theory, implementation, and application Probabilistic neural networks: a brief overview of theory...

    2020

  24. [32]

    \ Woolnough, S

    nerc2008cascade APACrefauthors Natural Environment Research Council , NERC CASCADE Project participants , Lister, G. \ Woolnough, S. APACrefauthors \ 2008 . Cascade -- Scale interactions in the tropical atmosphere model runs . Cascade -- Scale interactions in the tropical atmo...

    2008

  25. [33]

    , Bhojanapalli, S

    neyshabur2017exploringgeneralizationdeeplearning APACrefauthors Neyshabur, B. , Bhojanapalli, S. , McAllester, D. \ Srebro, N. APACrefauthors \ 2017 . Exploring Generalization in Deep Learning. Exploring generalization in deep learning. APACrefURL https://arxiv.org/abs/1706.08...

    2017 arXiv

  26. [34]

    , Fink, G A

    oberdiek2023uqganunifiedmodeluncertainty APACrefauthors Oberdiek, P. , Fink, G A. \ Rottmann, M. APACrefauthors \ 2023 . UQGAN: A Unified Model for Uncertainty Quantification of Deep Classifiers trained via Conditional GANs . UQGAN: A Unified Model for Uncertainty Quantificati...

    2023 arXiv

  27. [35]

    , Hogan, R J

    pearson2010 APACrefauthors Pearson, K J. , Hogan, R J. , Allan, R P. , Lister, G M S. \ Holloway, C E. APACrefauthors \ 2010 . Evaluation of the model representation of the evolution of convective systems using satellite observations of outgoing longwave radiation Evaluation o...

    2010 doi

  28. [36]

    , Torrecilla, J L

    ramos-carreno++_2024_scikit-fda APACrefauthors Ramos-Carreño, C. , Torrecilla, J L. , Carbajo Berrocal, M. , Marcos Manchón, P. \ Suárez, A. APACrefauthors \ 2024 05 . scikit-fda: A Python Package for Functional Data Analysis scikit-fda: A Python Package for Functional Data An...

    2024 doi

  29. [37]

    \ Wrede, P

    schneider1998artificial APACrefauthors Schneider, G. \ Wrede, P. APACrefauthors \ 1998 . Artificial neural networks for computer-based molecular design Artificial neural networks for computer-based molecular design . Progress in biophysics and molecular biology 70 3 175--222

    1998

  30. [38]

    APACrefauthors \ 2024 02

    ramos-carreno++_2024_scikit-fda-repo APACrefauthors scikit-fda developers, T. APACrefauthors \ 2024 02 . scikit-fda: Functional Data Analysis in Python. scikit-fda: Functional data analysis in python. APACrefURL https://github.com/GAA-UAM/scikit-fda APACrefURL APACrefDOI doi:1...

    2024 doi

  31. [39]

    APACrefauthors \ 1990

    specht1990probabilistic APACrefauthors Specht, D F. APACrefauthors \ 1990 . Probabilistic neural networks Probabilistic neural networks . Neural networks 3 1 109--118

    1990

  32. [40]

    APACrefauthors \ 2021

    stein2021parametric APACrefauthors Stein, M L. APACrefauthors \ 2021 . A parametric model for distributions with flexible behavior in both tails A parametric model for distributions with flexible behavior in both tails . Environmetrics 32 2 e2658

    2021

  33. [41]

    , Glaws, A

    stengel2020adversarial APACrefauthors Stengel, K. , Glaws, A. , Hettinger, D. \ King, R N. APACrefauthors \ 2020 . Adversarial super-resolution of climatological wind and solar data Adversarial super-resolution of climatological wind and solar data . Proceedings of the Nationa...

    2020

  34. [42]

    tian2020coarse APACrefauthors Tian, C. , Xu, Y. , Zuo, W. , Zhang, B. , Fei, L. \ Lin, C W. APACrefauthors \ 2020 . Coarse-to-fine CNN for image super-resolution Coarse-to-fine CNN for image super-resolution . IEEE Transactions on Multimedia 23 1489--1502

    2020

  35. [43]

    wang2018esrgan APACrefauthors Wang, X. , Yu, K. , Wu, S. , Gu, J. , Liu, Y. , Dong, C. Change Loy, C. APACrefauthors \ 2018 . Esrgan: Enhanced super-resolution generative adversarial networks Esrgan: Enhanced super-resolution generative adversarial networks . Proceedings of th...

    2018

  36. [44]

    , Chen, J

    wang2020deeplearningimagesuperresolution APACrefauthors Wang, Z. , Chen, J. \ Hoi, S C H. APACrefauthors \ 2020 . Deep Learning for Image Super-resolution: A Survey. Deep learning for image super-resolution: A survey. APACrefURL https://arxiv.org/abs/1902.06068 APACrefURL

    2020 arXiv

  37. [45]

    , Ren, W

    9700735 APACrefauthors Wen, W. , Ren, W. , Shi, Y. , Nie, Y. , Zhang, J. \ Cao, X. APACrefauthors \ 2022 . Video Super-Resolution via a Spatio-Temporal Alignment Network Video super-resolution via a spatio-temporal alignment network . IEEE Transactions on Image Processing 31 1...

    2022

  38. [46]

    yang2014single APACrefauthors Yang, C Y. , Ma, C. \ Yang, M H. APACrefauthors \ 2014 . Single-image super-resolution: A benchmark Single-image super-resolution: A benchmark . Computer Vision--ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proce...

    2014

Pith tools

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