Pith. sign in

REVIEW 3 major objections 4 minor 45 references

A Simple and Effective Method for Uncertainty Quantification and OOD Detection

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper's central claim is that an information potential field over a single network's feature vectors — a nonparametric density estimate — detects out-of-distribution images more accurately than existing baselines, while keeping…

desk verdict A clean but minor paper: KDE on features, with a promising 2D demo but a headline AUROC claim that is not credible because h is tuned on the test set. read the letter →

arxiv 2508.00754 v1 pith:F5JZ6F4C submitted 2025-08-01 cs.LG cs.AI

Yaxin Ma , Benjamin Colburn , Jose C. Principe This is my paper · ORCID
classification cs.LGcs.AI
keywords uncertaintyquantificationout-of-distributiondetectionkerneldensityestimationinformationpotentialfieldfeature-spacespectralnormalizationdistributionalshift
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

Bayesian and ensemble methods give uncertainty quantification but cost many forward passes and large storage. This paper proposes a deterministic alternative: train one network with spectral normalization, then treat the density of its top-layer feature vectors as the uncertainty signal, computed with a kernel-density field called the information potential field (IPF). A test point landing in a low-density region of that field is labeled out-of-distribution. On CIFAR-10 as in-distribution and SVHN as out-of-distribution, the paper reports AUROC 93.18 with WideResNet-28-10, slightly above DDU's 92.90 and DUQ's 92.43. The contribution is showing that this simple Parzen-style density estimate can match or beat more elaborate single-model uncertainty methods without per-class distributional assumptions.

What carries the argument

The information potential field (IPF) is the central object: a sum of isotropic Gaussian kernels centered at the training feature vectors, $\psi(z)=\frac{1}{N}\sum_i \exp(-\|z-z_i\|^2/2h^2)$, used as a nonparametric stand-in for the feature density $p(z)$. It is computed once after training, and at test time the scalar field value at $z^*$ is the OOD score. Its behavior is governed by the kernel width $h$, which the paper selects by cross-validation, and its usefulness depends on the distance-preserving feature map produced by spectral normalization.

What would settle it

Re-run the CIFAR-10/SVHN experiment with the kernel width chosen by Silverman's rule or by cross-validation on a held-out portion of the training set rather than by maximizing AUROC on the test set, and also try a second OOD set such as LSUN or Texture. If IPF no longer beats DDU and DUQ under honest bandwidth selection or generalizes to the second OOD set, the reported AUROC is an artifact of tuning rather than evidence that the density field is faithful.

Watch

Extended reading notes

Core claim

The central discovery is that a nonparametric density estimate of the feature space of a single deterministic classifier is enough to separate in-distribution from out-of-distribution inputs. The method defines $\psi(z)=\frac{1}{N}\sum_{i=1}^N G(z-z_i)$, where $G$ is an isotropic Gaussian kernel and the $z_i$ are the training-set feature embeddings from the layer before the classification head; low $\psi(z^*)$ for a test point $z^*$ means high uncertainty and an OOD decision. Spectral normalization is used during training to enforce a bi-Lipschitz constraint on the feature map, so distinct inputs do not collapse to the same feature and the density field remains meaningful. The paper claims this yields AUROC 93.18 on CIFAR-10 vs SVHN, above softmax, ensembles, DUQ, and DDU, and that the method is simpler because it models the whole feature space rather than one Gaussian per class.

Load-bearing premise

The whole comparison rests on the assumption that one Gaussian kernel width, applied in a 640-dimensional feature space, gives an honest estimate of where the training data live; the paper itself cautions that this style of density estimation is not expected to work above 20 dimensions.

Editorial extensions

If this is right

  • Out-of-distribution detection becomes a single forward pass plus density evaluation, so the computational and storage costs of ensembles and Bayesian approximations are avoided.
  • The density estimate makes no per-class Gaussian assumption, so it applies to feature spaces with overlapping or non-Gaussian class structure where DDU's mixture model is misspecified.
  • The kernel width $h$ acts as a single sensitivity knob, letting practitioners trade a tighter uncertainty region for broader tolerance.
  • For low-dimensional inputs, the same field can be evaluated directly in the data space, enabling fast OOD checks without any feature extractor.

Reading between the lines

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

  • The paper does not report how the 640-dimensional kernel width was validated beyond maximizing AUROC on the test distribution; an independent evaluation with bandwidth chosen by Silverman's rule or by an internal train/validation split would show whether the reported margin over DDU is robust or an artifact of tuning.
  • Because the paper itself notes that Parzen estimation is not recommended above 20 dimensions, the strong result in 640 dimensions is surprising; replacing the isotropic Gaussian with a dimension-adaptive or infinitely divisible kernel density estimator, which the paper names as future work, is a direct test of whether the IPF principle or the simple kernel is what carries the performance.
  • The IPF assigns a scalar uncertainty to every input without retraining, so the same score could in principle be used for active-learning acquisition and for OOD detection on other benchmarks; neither application is tested here.
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 / 4 minor

Summary. The paper proposes Information Potential Field (IPF), a kernel density estimate over the penultimate-layer features of a single deterministic neural network, as a method for uncertainty quantification and out-of-distribution (OOD) detection. The method is evaluated on two synthetic 2D datasets (Two Moons, Three Spirals) with visual uncertainty maps, and on CIFAR-10 vs. SVHN OOD detection using a WideResNet-28-10, reporting an AUROC of 93.18 for IPF versus 92.90 for DDU. The authors claim that IPF outperforms baseline methods while being simpler and requiring only a single forward pass.

Significance. If the reported result were robust, the paper would offer a simple, appealing alternative to class-conditional Gaussian density estimation (DDU) and distance-based methods, with no additional training cost at inference. The synthetic experiments are visually clear and illustrate the method's behavior in low dimensions. The paper also honestly identifies the main limitation of Parzen estimation in high dimensions and points to more advanced estimators as future work. However, the empirical case for the central claim is weakened by how the kernel-width hyperparameter is selected and by the application of the method in a 640-dimensional feature space, which the paper itself says is beyond the reliable regime for Parzen estimation. These issues currently limit the significance of the contribution.

major comments (3)
  1. [Section III-C, Table I] The kernel width h is selected by "performing cross-validation over the range [0.01, 1], selecting the value that maximized the AUROC score." Because AUROC is computed on the CIFAR-10 test set versus the SVHN OOD set, this procedure fits h directly to the test OOD pair. The reported AUROC of 93.18 is therefore an in-sample, optimistic estimate rather than an independent performance measure. Since the margin over DDU is only 0.28 AUROC, the central claim that IPF outperforms baselines is not supported unless h is selected on a separate validation set (or via nested cross-validation) and the AUROC is then reported on a held-out test pair.
  2. [Section IV and Section III-C] The paper states that "Parzen estimation does not scale well, and in practice it should not be used above 20 dimensions," yet the method applies an isotropic Gaussian KDE in the 640-dimensional feature space of WideResNet-28-10. This contradiction undermines the reliability of the density estimates underlying the reported AUROC. The claim that "we were surprised with the quality of the Parzen estimator" is not evidence. The authors should provide a sanity check of the density estimates in this setting, for example by comparing ID and OOD density score distributions, testing sensitivity to h across a wider range, or comparing against a higher-dimensional density estimator, before the AUROC numbers can be interpreted as evidence of a sound method.
  3. [Section III-C, Table I] The OOD evaluation consists of a single dataset pair (CIFAR-10 as ID, SVHN as OOD) with no significance test and no variation of the in-distribution dataset or model architecture. The abstract and discussion claim that "the results demonstrate that our method outperforms baseline models" and that IPF "can improve state of the art results for OOD detection," but a single pair with a 0.28 AUROC difference and overlapping standard deviations (93.18 ± 0.006 versus 92.90 ± 0.016) is insufficient to support a general superiority claim. The authors should either broaden the evaluation (e.g., multiple OOD datasets, multiple architectures) or temper the claims accordingly.
minor comments (4)
  1. [Section II-C, Eq. (3)] Equation (3) omits the normalization constant of the Gaussian kernel, so ψ(z) in Eq. (2) is not a probability density but an unnormalized kernel sum. Since AUROC is rank-based this does not affect the reported metric, but the text should avoid calling ψ(z) an estimate of p(z) without noting the missing constant.
  2. [Section II-B] There is a typo in "to improve the quality of the the features" in the paragraph after Eq. (1); "the the" should be "the."
  3. [Section III-B] The choice of the best kernel size (0.3) for the synthetic experiments is reported without a selection criterion; stating whether this was chosen visually or by a quantitative rule would improve reproducibility.
  4. [Section IV] The phrase "willl be pursued" contains a typo; it should read "will be pursued."

Circularity Check

1 steps flagged · score 6.0 of 10

AUROC superiority is partly a fitted statistic: kernel width h is selected by maximizing AUROC on the same CIFAR-10/SVHN test pair used to report Table I.

  1. fitted input called prediction [Section III-C, Eq. (3), and Table I]
    "To determine the optimal kernel width, we performed cross-validation over the range [0.01, 1], selecting the value that maximized the AUROC score. The best kernel width selected was 0.35."

    The only free parameter of the detector, h in the isotropic Gaussian kernel (Eq. 3), is selected by maximizing AUROC, which is exactly the metric and OOD test pair (SVHN vs CIFAR-10) later reported in Table I as 'Our method (IPF) 93.18'. The reported number is therefore the value of the objective used to choose h, not an independent prediction. The margin over DDU (92.90) is only 0.28 AUROC, small enough to be entirely an artifact of this in-sample hyperparameter optimization. No separate validation split or nested OOD-label-free selection is described, so the headline result is statistically forced by the selection criterion rather than demonstrated by held-out evaluation.

full rationale

The paper has no mathematical derivation chain to be circular: Eq. (2) is a definition of a kernel density estimator (Parzen/IPF), and the method is applied as an algorithm. The circularity is localized to evaluation. Section III-C reports that h was selected by maximizing AUROC over [0.01,1], and Table I reports AUROC as evidence of superiority. Since AUROC on the CIFAR-10/SVHN pair is both the selection objective and the reported metric, the central empirical claim ('outperforms baseline models') rests on an optimized in-sample statistic. The paper's own admission that Parzen estimation should not be used above 20 dimensions (Section IV, citing [37]) is a correctness concern, not a circular step. The citation [37] to the co-author's book is used to motivate IPF and state its limits, but the method itself is standard KDE and does not depend on that citation for its content, so it is not load-bearing. I flag one strong circular step (fitted-input-as-prediction) and therefore score 6 rather than higher or lower. If a separate validation procedure for h is later provided, the empirical claim would be properly supported.

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

The only tuned quantity is the kernel width h, and it is selected using the evaluation metric. The 'information potential field' is a known kernel density estimator from prior literature (Principe 2010), not a newly postulated entity. The other assumptions are standard KDE and spectral normalization premises, both flagged as fragile in the paper's own discussion.

free parameters (2)
  • Kernel width h (synthetic 2D tasks) = 0.3
    Selected from range 0.1 to 1 as 'the best kernel size' based on visual uncertainty maps, not held-out validation; affects the uncertainty regions in Fig. 1 and 2.
  • Kernel width h (CIFAR-10 vs SVHN) = 0.35
    Selected by 'cross-validation over the range [0.01, 1], selecting the value that maximized the AUROC score' (Section III-C). Because AUROC is the evaluation metric, this is fitting the reported result to the test data.
assumptions (3)
  • domain assumption The information potential field ψ(z) approximates the feature space density p(z) (Eq. 2 and surrounding text).
    KDE is a consistent density estimator only under regularity conditions and as bandwidth tends to 0 with sample size; in 640 dimensions with a single isotropic bandwidth this approximation is fragile, as the paper concedes in Section IV.
  • domain assumption Spectral Normalization enforces the bi-Lipschitz constraint and prevents feature collapse (Section II-B, citing [30]).
    The method inherits the assumption that SN makes the feature map distance-preserving enough for density and distance measurements to be meaningful.
  • ad hoc to paper Isotropic Gaussian kernel is an appropriate choice for the feature space.
    The paper chooses the isotropic Gaussian 'due to its simplicity' (Section IV), without evidence that feature covariances are isotropic; this choice directly affects OOD decision boundaries.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Simple and Effective Method for Uncertainty Quantification and OOD Detection." pith.science (2026). https://pith.science/paper/F5JZ6F4C

@misc{pith2026250800754,
  author       = {Pith},
  title        = {Pith review of: A Simple and Effective Method for Uncertainty Quantification and OOD Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F5JZ6F4C}},
  note         = {Machine review of arXiv:2508.00754}
}
read the original abstract

Bayesian neural networks and deep ensemble methods have been proposed for uncertainty quantification; however, they are computationally intensive and require large storage. By utilizing a single deterministic model, we can solve the above issue. We propose an effective method based on feature space density to quantify uncertainty for distributional shifts and out-of-distribution (OOD) detection. Specifically, we leverage the information potential field derived from kernel density estimation to approximate the feature space density of the training set. By comparing this density with the feature space representation of test samples, we can effectively determine whether a distributional shift has occurred. Experiments were conducted on a 2D synthetic dataset (Two Moons and Three Spirals) as well as an OOD detection task (CIFAR-10 vs. SVHN). The results demonstrate that our method outperforms baseline models.

Figures

Figures reproduced from arXiv: 2508.00754 by the authors.

Figure 1
Figure 1. Uncertainty results of different baseline methods on the Two-Moons dataset and Three-Spirals dataset. The first row corresponds to the Two-Moons [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Uncertainty results on the Two-Moons and Three-spirals dataset based on different kernel size. The first row corresponds to the Two-Moons dataset, [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Uncertainty results for the Two-Moons dataset and Three-Spirals [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 38 canonical work pages

  1. [1]

    Deep learning,

    Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436-444, May 2015

    work page 2015

  2. [2]

    ImageNet classification with deep convolutional neural networks,

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” Adv. Neural Inf. Process. Syst., vol. 25, pp. 1097-1105, 2012

    work page 2012

  3. [3]

    A survey of the usages of deep learning for natural language processing,

    D. W. Otter, J. R. Medina, and J. K. Kalita, “A survey of the usages of deep learning for natural language processing,” IEEE Trans. Neural Netw. Learn. Syst., vol. 32, no. 2, pp. 604-624, Feb. 2020

    work page 2020

  4. [4]

    A survey of deep learning techniques for autonomous driving,

    S. Grigorescu, B. Trasnea, T. Cocias, and G. Macesanu, “A survey of deep learning techniques for autonomous driving,” J. Field Robot., vol. 37, no. 3, pp. 362-386, Mar. 2020

    work page 2020

  5. [5]

    Human breast numerical model generation based on deep learning for photoacoustic imaging,

    Y . Ma, C. Yang, J. Zhang, Y . Wang, F. Gao, and F. Gao, “Human breast numerical model generation based on deep learning for photoacoustic imaging,” in *Proc. IEEE Eng. Med. Biol. Soc. (EMBC)*, Jul. 2020, pp. 1919-1922

    work page 2020

  6. [6]

    BPEN: Brain Posterior Evidential Network for trustworthy brain imaging analysis,

    K. Ye, H. Tang, S. Dai, I. Fortel, P. M. Thompson, R. S. Mackin, A. Leow, H. Huang, L. Zhan, and Alzheimer’s Disease Neuroimaging Initiative, “BPEN: Brain Posterior Evidential Network for trustworthy brain imaging analysis,” Neural Netw., vol. 183, p. 106943, 2025

    work page 2025

  7. [7]

    Defining Boundaries: A Spectrum of Task Feasibility for Large Language Models,

    W. Zhang, Z. Xu, and H. Cai, “Defining Boundaries: A Spectrum of Task Feasibility for Large Language Models,” *arXiv preprint arXiv:2408.05873*, 2024

    arXiv 2024

  8. [8]

    A review of uncertainty quantification in deep learning: Techniques, applications and challenges,

    M. Abdar *et al.*, “A review of uncertainty quantification in deep learning: Techniques, applications and challenges,” Inf. Fusion, vol. 76, pp. 243-297, 2021

    work page 2021

Show all 45 references
  1. [9]

    Aleatory or epistemic? Does it matter?

    A. Der Kiureghian and O. Ditlevsen, “Aleatory or epistemic? Does it matter?” Struct. Saf., vol. 31, no. 2, pp. 105-112, 2009

    2009

  2. [10]

    Aleatoric and epistemic uncertainty in machine learning: An introduction to concepts and methods,

    E. H ¨ullermeier and W. Waegeman, “Aleatoric and epistemic uncertainty in machine learning: An introduction to concepts and methods,” Mach. Learn., vol. 110, no. 3, pp. 457-506, 2021

    2021

  3. [11]

    A survey on un- certainty quantification methods for deep learning,

    W. He, Z. Jiang, T. Xiao, Z. Xu, and Y . Li, “A survey on un- certainty quantification methods for deep learning,” *arXiv preprint arXiv:2302.13425*, 2023

    2023

  4. [12]

    A survey of uncertainty in deep neural networks,

    J. Gawlikowski *et al.*, “A survey of uncertainty in deep neural networks,” Artif. Intell. Rev., vol. 56, Suppl. 1, pp. 1513-1589, 2023

    2023

  5. [13]

    Active learning with statistical models,

    D. A. Cohn, Z. Ghahramani, and M. I. Jordan, “Active learning with statistical models,” J. Artif. Intell. Res., vol. 4, pp. 129-145, 1996

    1996

  6. [14]

    Active learning for convolutional neural networks: A core-set approach,

    O. Sener and S. Savarese, “Active learning for convolutional neural networks: A core-set approach,” *arXiv preprint arXiv:1708.00489*, 2017

    2017 arXiv

  7. [15]

    Deep Bayesian Active Learning with Image Data,

    Y . Gal, R. Islam, and Z. Ghahramani, “Deep Bayesian Active Learning with Image Data,” in *Proc. 34th Int. Conf. Mach. Learn.*, 2017, pp. 1183-1192

    2017

  8. [16]

    Generalized ODIN: Detecting out-of-distribution image without learning from out-of-distribution data,

    Y . C. Hsu, Y . Shen, H. Jin, and Z. Kira, “Generalized ODIN: Detecting out-of-distribution image without learning from out-of-distribution data,” in *Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit.*, 2020, pp. 10951-10960

    2020

  9. [18]

    R. M. Neal, *Bayesian Learning for Neural Networks*, Ph.D. disserta- tion, Univ. Toronto, 1995

    1995

  10. [19]

    Bayesian training of backpropagation networks by the hybrid Monte Carlo method,

    R. M. Neal, “Bayesian training of backpropagation networks by the hybrid Monte Carlo method,” Tech. Rep. CRG-TR-92-1, Dept. Comput. Sci., Univ. Toronto, 1992

    1992

  11. [20]

    Transforming neural-net output levels to probability distributions,

    J. Denker and Y . LeCun, “Transforming neural-net output levels to probability distributions,” Adv. Neural Inf. Process. Syst., vol. 3, 1990

    1990

  12. [21]

    A practical Bayesian framework for backpropagation networks,

    D. J. MacKay, “A practical Bayesian framework for backpropagation networks,” Neural Comput., vol. 4, no. 3, pp. 448-472, 1992

    1992

  13. [22]

    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,” in *Proc. Int. Conf. Mach. Learn.*, 2016, pp. 1050-1059

    2016

  14. [23]

    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,” Adv. Neural Inf. Process. Syst., vol. 30, 2017

    2017

  15. [24]

    Evidential deep learning to quantify classification uncertainty,

    M. Sensoy, L. Kaplan, and M. Kandemir, “Evidential deep learning to quantify classification uncertainty,” Adv. Neural Inf. Process. Syst., vol. 31, 2018

    2018

  16. [25]

    Predictive uncertainty estimation via prior networks,

    A. Malinin and M. Gales, “Predictive uncertainty estimation via prior networks,” Adv. Neural Inf. Process. Syst., vol. 31, 2018

    2018

  17. [26]

    A simple approach to improve single-model deep uncertainty via distance-awareness,

    J. Z. Liu *et al.*, “A simple approach to improve single-model deep uncertainty via distance-awareness,” J. Mach. Learn. Res., vol. 24, no. 42, pp. 1-63, 2023

    2023

  18. [27]

    Density-softmax: Efficient test-time model for uncertainty estimation and robustness under distribution shifts,

    H. M. Bui and A. Liu, “Density-softmax: Efficient test-time model for uncertainty estimation and robustness under distribution shifts,” in *Proc. 41st Int. Conf. Mach. Learn.*, Jul. 2024, pp. 4822-4853

    2024

  19. [28]

    Discriminant Distance-Aware Rep- resentation on Deterministic Uncertainty Quantification Methods,

    J. Zhang, K. Das, and S. Kumar, “Discriminant Distance-Aware Rep- resentation on Deterministic Uncertainty Quantification Methods,” in *Proc. Int. Conf. Artif. Intell. Stat.*, Apr. 2024, pp. 2917-2925

    2024

  20. [29]

    Uncertainty estimation using a single deep deterministic neural network,

    J. van Amersfoort, L. Smith, Y . W. Teh, and Y . Gal, “Uncertainty estimation using a single deep deterministic neural network,” in *Proc. Int. Conf. Mach. Learn.*, Nov. 2020, pp. 9690-9700

    2020

  21. [30]

    Simple and principled uncertainty estimation with deterministic deep learning via distance awareness,

    J. Liu, Z. Lin, S. Padhy, D. Tran, T. B. Weiss, and B. Lakshminarayanan, “Simple and principled uncertainty estimation with deterministic deep learning via distance awareness,” Adv. Neural Inf. Process. Syst., vol. 33, pp. 7498-7512, 2020

    2020

  22. [31]

    Deep deterministic uncertainty: A new simple baseline,

    J. Mukhoti, A. Kirsch, J. van Amersfoort, P. H. Torr, and Y . Gal, “Deep deterministic uncertainty: A new simple baseline,” in *Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit.*, 2023, pp. 24384-24394

    2023

  23. [32]

    A baseline for detecting misclassified and out-of-distribution examples in neural networks,

    D. Hendrycks and K. Gimpel, “A baseline for detecting misclassified and out-of-distribution examples in neural networks,” *arXiv preprint arXiv:1610.02136*, 2016

    2016 arXiv

  24. [33]

    Enhancing the reliability of out- of-distribution image detection in neural networks,

    S. Liang, Y . Li, and R. Srikant, “Enhancing the reliability of out- of-distribution image detection in neural networks,” *arXiv preprint arXiv:1706.02690*, 2017

    2017 arXiv

  25. [34]

    A simple unified framework for detecting out-of-distribution samples and adversarial attacks,

    K. Lee, K. Lee, H. Lee, and J. Shin, “A simple unified framework for detecting out-of-distribution samples and adversarial attacks,” Adv. Neural Inf. Process. Syst., vol. 31, 2018

    2018

  26. [35]

    Energy-based out-of-distribution detection,

    W. Liu, X. Wang, J. Owens, and Y . Li, “Energy-based out-of-distribution detection,” Adv. Neural Inf. Process. Syst., vol. 33, pp. 21464-21475, 2020

    2020

  27. [36]

    On feature collapse and deep kernel learning for single forward pass uncertainty,

    J. van Amersfoort, L. Smith, A. Jesson, O. Key, and Y . Gal, “On feature collapse and deep kernel learning for single forward pass uncertainty,” *arXiv preprint arXiv:2102.11409*, 2021

    2021 arXiv

  28. [37]

    J. C. Principe, *Information Theoretic Learning: Renyi’s Entropy and Kernel Perspectives*, Springer Sci. and Bus. Media, 2010, pp. 57-58

    2010

  29. [38]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky and G. Hinton, “Learning multiple layers of features from tiny images,” Univ. Toronto, 2009

    2009

  30. [39]

    Reading digits in natural images with unsupervised feature learning,

    Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, and A. Y . Ng, “Reading digits in natural images with unsupervised feature learning,” in *Proc. NIPS Workshop Deep Learn. Unsupervised Feature Learn.*, Dec. 2011, vol. 2011, no. 2, p. 4

    2011

  31. [40]

    B. W. Silverman, *Density Estimation for Statistics and Data Analysis*, Chapman and Hall/CRC, 1986

    1986

  32. [41]

    Obtaining well calibrated probabilities using Bayesian binning,

    M. P. Naeini, G. Cooper, and M. Hauskrecht, “Obtaining well calibrated probabilities using Bayesian binning,” in *Proc. AAAI Conf. Artif. Intell.*, Feb. 2015, vol. 29, no. 1

    2015

  33. [42]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in *Proc. IEEE Conf. Comput. Vis. Pattern Recognit.*, 2016, pp. 770-778

    2016

  34. [43]

    Wide residual networks,

    S. Zagoruyko, “Wide residual networks,” *arXiv preprint arXiv:1605.07146*, 2016

    2016 arXiv

  35. [44]

    Visualizing data using t-SNE,

    L. van der Maaten and G. Hinton, “Visualizing data using t-SNE,” J. Mach. Learn. Res., vol. 9, pp. 2579-2605, 2008

    2008

  36. [45]

    Measures of entropy from data using infinitely divisible kernels,

    L. G. S. Giraldo, M. Rao, and J. C. Principe, “Measures of entropy from data using infinitely divisible kernels,” IEEE Trans. Inf. Theory, vol. 61, no. 1, pp. 535-548, 2014

    2014

  37. [46]

    Understanding autoencoders with information theoretic concepts,

    S. Yu and J. C. Principe, “Understanding autoencoders with information theoretic concepts,” Neural Netw., vol. 117, pp. 104-123, 2019

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.