Pith. sign in

REVIEW 5 major objections 7 minor 1 cited by

Epistemic Wrapping for Uncertainty Quantification

T0 review · 5 major / 7 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read The paper claims that converting a Bayesian neural network's weight posteriors into belief-function posteriors and then running inference through a hybrid interval network improves classification accuracy and out-of-distribution detection…

desk verdict Novel parameter-space belief-function wrapper idea is undone by Section 3.6, where intervals are collapsed to midpoints and uncertainty comes from MC Dropout, so the claimed mechanism is never actually tested. read the letter →

arxiv 2505.02277 v2 pith:QQNV45FF submitted 2025-05-04 cs.LG

classification cs.LG
keywords epistemicuncertaintybelieffunctionsBayesianneuralnetworksintervalquantificationout-of-distributiondetectionDirichletdistribution
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

This paper tries to establish that a post hoc transformation of a Bayesian neural network's weight posteriors into belief-function posteriors improves both classification accuracy and uncertainty quantification. The transformation, called the Epistemic Wrapper, is meant to work for any BNN without retraining it: only a small budgeted subset of weights is converted, and inference is then run through a hybrid interval network. On MNIST the wrapped model reaches 91.02% accuracy versus 72.44% for the baseline BNN, and on Fashion-MNIST it reaches 82.45% versus 58.91%. Out-of-distribution detection also improves, with AUROC rising from 0.5329 to 0.6673 when Fashion-MNIST is used as the OOD set for an MNIST-trained model. A sympathetic reader would care because the method promises a general, cheap upgrade to uncertainty-aware classification from an already-trained Bayesian posterior.

What carries the argument

The object that carries the argument is a five-step transformation pipeline applied to the BNN's Gaussian weight posteriors. First the posteriors are truncated to dynamic bounds; second, a continuous belief function is computed over closed intervals, with plausibility defined as the supremum of the normalised posterior on the interval; third, Möbius inversion turns belief values into mass values; fourth, a Dirichlet distribution is fitted to the mass grid by the method of L-moments; fifth, the resulting belief posteriors are turned into weight intervals that initialise a Hybrid Interval Neural Network, which averages upper and lower interval bounds during inference. The theoretical anchor is the containment relation $\mathrm{Bel}(A)\leq P(A)\leq \mathrm{Pl}(A)$, which the paper uses to argue that the wrapper is conservative and consistent with Bayesian inference.

What would settle it

Hold everything fixed but replace the Dirichlet-derived interval bounds with equal-width intervals sampled directly from the BNN posterior; if accuracy and out-of-distribution AUROC remain at the wrapped values, the belief-function conversion is not the cause of the gains.

Watch

Extended reading notes

Core claim

The paper's central claim is that epistemic uncertainty can be modelled directly in the parameter space of a neural network by wrapping a learned Bayesian posterior into a belief-function posterior, and that doing so yields better generalisation and uncertainty estimation. The construction is conservative: for every measurable set $A$ of parameters, the original posterior probability $P(A)$ lies between the transformed belief $\mathrm{Bel}(A)$ and plausibility $\mathrm{Pl}(A)$, so the wrapping adds second-order uncertainty without discarding the Bayesian information. Inference is performed by a Hybrid Interval Neural Network whose weight intervals come from Dirichlet-derived samples for wrapped parameters and Gaussian intervals for the rest. The authors report that the wrapper beats both the variational BNN baseline and a randomly initialised interval network, before and after fine-tuning, and that it assigns higher epistemic uncertainty to out-of-distribution inputs.

Load-bearing premise

The paper's load-bearing step is the final conversion of the fitted uncertainty distribution into interval bounds on the network's weights; it is described only in words, with no formula, so if that mapping is arbitrary or lossy the reported gains could come from fine-tuning the interval network rather than from the wrapping itself.

Editorial extensions

If this is right

  • Any variational BNN with Gaussian priors can in principle be wrapped after training, with only about 5% of weights transformed, to improve test accuracy: 91.02% vs 72.44% on MNIST and 82.45% vs 58.91% on Fashion-MNIST after fine-tuning.
  • Out-of-distribution detection improves: on Fashion-MNIST images presented to an MNIST-trained model, the wrapped interval network raises AUROC from 0.5329 to 0.6673 and roughly doubles the epistemic-uncertainty score on OOD inputs.
  • The wrapper is conservative in the sense that the original Bayesian posterior is contained in the credal set defined by the transformed belief function, so the method preserves Bayesian predictive information while enriching it with second-order uncertainty.
  • Budgeting matters: selecting the 5% of weights with high mean or high variance outperforms random selection, so the benefit comes from wrapping informative or uncertain parameters rather than any subset.
  • Because the BNN itself is not retrained and only the initialisation of the Hybrid-INN changes, the procedure is a drop-in uncertainty-quantification upgrade compatible with existing variational BNNs.

Reading between the lines

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

  • If the conservativeness property holds for the actual weight intervals delivered to the network, one testable extension is that predictive rejection curves and OOD scores should remain stable when the number of closed intervals is varied; the appendix shows interval-count sensitivity but the main text fixes it at 30.
  • The method's reliance on a small, budgeted subset of weights suggests a possible link to sparse or pruning-based Bayesian inference: the wrapper could double as an importance measure for deciding which parameters most need second-order treatment.
  • A stronger OOD evaluation would pair MNIST with random noise or a structurally different digit-like corpus rather than Fashion-MNIST, which shares low-level features with MNIST; that would separate genuine out-of-distribution sensitivity from feature-level differences.
  • The claimed generality across 'any BNN baseline' is not yet demonstrated beyond single-hidden-layer MLPs on two datasets; applying the wrapper to convolutional Bayesian backbones would settle whether the parameter-space belief wrapping scales.
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

5 major / 7 minor

Summary. The paper introduces 'Epistemic Wrapping,' a post hoc procedure that converts the weight posteriors of a trained Bayesian neural network into belief-function posteriors over parameter intervals. The pipeline consists of five steps: extracting the BNN posterior, truncating each univariate posterior, computing continuous belief functions on a grid of closed intervals via Moebius inversion, fitting a Dirichlet distribution to the resulting masses by a method the authors call weighted L-moments, and performing inference with a Hybrid Interval Neural Network whose weights are derived from the wrapped and unwrapped posteriors. Experiments on MNIST and Fashion-MNIST compare the Epi-Wrapper against a randomly initialized INN and a BNN baseline, before and after fine-tuning, and report classification accuracy plus OOD detection metrics (AUROC, AUPRC, epistemic uncertainty). The abstract claims comprehensive experiments on MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100, but the body only reports MNIST and Fashion-MNIST results.

Significance. If the central mechanism worked as advertised, the paper would make a useful methodological contribution: it is among the few attempts to model epistemic uncertainty in parameter space with higher-order measures, and it proposes a transformation that is, in principle, applicable to any BNN posterior without retraining. The paper also has some praiseworthy elements: it reports means and standard deviations over 15 runs, includes ablation studies on budgeting strategies and on the number of intervals, and is candid about the Hybrid-INN's interval-averaging rule. However, as detailed in the major comments, the evaluation does not actually exercise the belief-function posterior at inference time, the key interval-mapping step is unspecified, and the after-fine-tuning accuracy results do not support the claim of significant generalization gains. The significance of the claimed contribution is therefore not established by the present manuscript.

major comments (5)
  1. [Section 3.6 and Section 4.4] The paper states that 'the way we handle these intervals is by computing the mean of the upper and lower bounds' and that predictive uncertainty is estimated with MC Dropout in Section 4.4. This means that during inference all wrapped intervals are collapsed to their midpoints, and no interval arithmetic, no belief/plausibility computation, and no Dirichlet sampling is used to obtain the reported predictions or uncertainty scores. Consequently, the OOD detection improvements in Table 3 (AUROC 0.6673 vs 0.5329) cannot be attributed to the epistemic wrapping mechanism, as the same results could be generated by the fine-tuned initialization alone with MC Dropout. The central claim that the Epistemic Wrapper 'significantly enhances ... uncertainty quantification' is therefore unsupported.
  2. [Section 3.6] The conversion from samples of the fitted Dirichlet distribution over interval masses to lower and upper bounds on the Hybrid-INN weights is never given as a formula or algorithm. The text only says that 'weight intervals are derived from a combination of Dirichlet-derived intervals and Gaussian posteriors.' Without this specification, the experiments are not reproducible, and the reader cannot verify that the implemented model corresponds to the belief-function posterior described in Sections 3.3 and 3.4. This is a load-bearing gap because the paper's entire contribution rests on that mapping.
  3. [Table 2 and Section 4.5] The claim that the Epi-Wrapper 'significantly outperforms' the baseline is contradicted by the after-fine-tuning numbers. On MNIST the Epi-Wrapper achieves 91.02 ± 0.05 against the INN's 91.12 ± 0.08, which is a slight decrease, and on Fashion-MNIST the values are 82.45 ± 0.10 vs 82.41 ± 0.19, which are statistically indistinguishable. The dramatic before-fine-tuning advantage (51.33% vs 9.33% on MNIST) compares a network initialized from a trained BNN posterior against a randomly initialized INN; this is an initialization comparison, not a test of the belief-function representation.
  4. [Abstract and Section 4.1] The abstract claims 'Comprehensive experiments ... on the MNIST, Fashion-MNIST, CIFAR-10 and CIFAR-100 datasets,' but no experimental results for CIFAR-10 or CIFAR-100 appear in the paper. Section 4.1 lists CIFAR-10 in the dataset description, yet all reported experiments use MNIST and Fashion-MNIST only; CIFAR-100 appears only in an appendix dataset description. The abstract's claim is therefore inaccurate.
  5. [Section 3.4, Theoretical Properties] The asserted inequality Bel(A) ≤ P(A) ≤ Pl(A) for all measurable A is stated without proof. With Pl defined as the supremum of the normalized posterior density over A and Bel(A) = 1 − Pl(A^c), it is not generally true that the probability measure P lies between Bel and Pl; a normalized density can exceed 1, and the complement operation can produce negative belief values. A rigorous derivation or a corrected statement is needed before this property can be cited as support for the method.
minor comments (7)
  1. [Section 3.6] For the unwrapped weights the interval is defined as Lower bound = μ − σ and Upper bound = μ + σ, but for the wrapped weights no analogous formula is given; please specify how the 'Dirichlet-derived intervals' are translated into numeric lower and upper bounds, and how the midpoint averaging is implemented in the forward pass.
  2. [Table 3] The iD Test Accuracy values in Table 3 (91.07 ± 0.08 for INN and 91.12 ± 0.06 for Epi-Wrapper) are inconsistent with the after-fine-tuning values in Table 2 (91.12 ± 0.08 and 91.02 ± 0.05) for what appears to be the same experimental setting; please reconcile the two tables.
  3. [Appendix A.3] The sentence 'The results before fine-tuning are presented in Table ??' references a missing table; either include the table or remove the reference.
  4. [Section 4.2] The text says 'MLP size (no hidden units) = 8' while the tables use 'MLP size' to mean the number of hidden units; this wording is confusing and should be clarified.
  5. [References] References [45] and [46] both cite Shafer's 'A mathematical theory of evidence' and appear to be duplicates; also, the citation style for [53] is inconsistent with the surrounding references.
  6. [Conclusions] The word 'Fahion-MNIST' in the concluding paragraph is a typo for 'Fashion-MNIST'.
  7. [Algorithm 1] The procedure called 'Weighted L-Moments' uses weighted mean and variance with weights given by the mass values, which is not the standard L-moments methodology of Hosking; please justify this estimator or rename it to avoid confusion.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity; the main issues are implementation/evaluation mismatches, not a derivation reducing to its inputs.

full rationale

The claimed derivation chain is not circular in the sense this pass targets. The wrapper starts from an independently trained BNN posterior and applies a fixed transformation: Pl and Bel are defined from the posterior density in Eqs. (3)-(4), masses are obtained by Moebius inversion, and a Dirichlet distribution is fitted to those masses by L-moments. No headline quantity (test accuracy, AUROC, AUPRC, EU) is used as an input to that transformation, and no fitted parameter is renamed as a prediction. The theoretical-property paragraph cites [9] (an author's book), but the stated Bel(A) <= P(A) <= Pl(A) sandwich follows directly from the construction, so the self-citation is not load-bearing. The paper's serious weaknesses are validity problems rather than circularity: Section 3.6 states 'the way we handle these intervals is by computing the mean of the upper and lower bounds', so the interval/belief structure is not propagated to the output; Section 4.4 estimates epistemic uncertainty with MC Dropout rather than from the belief posterior; and Section 4.3 fine-tunes on the same training set, which confounds the comparison with the 72.44% BNN number. Those problems undermine attribution of the reported gains, but they do not exhibit an equation that turns the wrapper's inputs into its claimed predictions by construction.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The method rests on five domain-specific choices: the variational posterior is trusted as the epistemic source; Wasserman's transformation is assumed valid without stated conditions; grid discretization is assumed lossless enough; L-moment fitting is assumed to produce a meaningful Dirichlet; and the unspecified Dirichlet-to-interval conversion is assumed faithful. The last two are ad hoc to this paper, and each is load-bearing for the claimed uncertainty improvement.

free parameters (5)
  • dynamic_multiplier = min(5.0, 1/sigma)
    Hand-chosen truncation rule in Section 3.2; controls how much of each weight posterior is kept and is not derived from data or theory.
  • number_of_closed_intervals = 30
    Fixed hyperparameter in Section 4.1; the appendix varies it, but all main results use 30.
  • posterior_sample_count = 5,000
    Fixed hyperparameter in Section 4.1 controlling how many BNN weight posterior samples are drawn.
  • budget_percentage = 5%
    Fixed hyperparameter in Section 4.1; the appendix shows 10-20% can be better after fine-tuning, so 5% is a choice that favors the reported story.
  • dirichlet_alpha_per_weight = not reported
    Estimated per weight by Eq. (8) from weighted L-moments of the mass grid; these fitted values determine the sampled interval weights used at inference.
assumptions (5)
  • domain assumption The variational posterior q(omega) from a DenseFlipout BNN is a faithful representation of parameter uncertainty that can be wrapped without losing information.
    Section 3.1 treats the learned posterior as the baseline whose transformation is claimed to enrich uncertainty; no check of variational approximation quality is provided.
  • standard math Wasserman's likelihood-based transformation (Eqs. 3-4) yields a valid belief function satisfying Bel(A) <= P(A) <= Pl(A) for all measurable sets A.
    Invoked from refs [53,9] in Sections 3.3 and 3.4; the normalization and bounded-domain conditions needed for the inequality across all A are not stated.
  • domain assumption Discretizing the posterior onto a grid of 30 closed intervals and applying Moebius inversion recovers the continuous mass function closely enough.
    Section 3.3 caps complexity by sampling belief values on a grid; no error bound, convergence analysis, or grid-sensitivity result is given in the main text.
  • ad hoc to paper Weighted L-moments over interval coordinates estimate a valid Dirichlet distribution whose samples represent the belief posterior.
    Eq. (8) relates L1/L2 of 3D simplex coordinates to alpha; no derivation connects weighted interval coordinates to Dirichlet samples, and the formula is used solely to make the wrapping pipeline operational.
  • ad hoc to paper Sampling from the fitted Dirichlet and converting the samples to interval weights for a Hybrid-INN preserves epistemic uncertainty.
    Section 3.6 states intervals are derived from Dirichlet-derived intervals and Gaussian posteriors but never defines the conversion; the entire inference claim rests on this step.
invented entities (1)
  • Hybrid Interval Neural Network (Hybrid-INN)
    purpose: Inference architecture that averages upper and lower bounds of wrapped and unwrapped interval weights, replacing standard interval arithmetic with mean bounds.
    New model variant introduced in Section 3.6; no external benchmark or theoretical result supports it beyond the paper's own comparisons.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Epistemic Wrapping for Uncertainty Quantification." pith.science (2026). https://pith.science/paper/QQNV45FF

@misc{pith2026250502277,
  author       = {Pith},
  title        = {Pith review of: Epistemic Wrapping for Uncertainty Quantification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QQNV45FF}},
  note         = {Machine review of arXiv:2505.02277}
}
read the original abstract

Uncertainty estimation is pivotal in machine learning, especially for classification tasks, as it improves the robustness and reliability of models. We introduce a novel `Epistemic Wrapping' methodology aimed at improving uncertainty estimation in classification. Our approach uses Bayesian Neural Networks (BNNs) as a baseline and transforms their outputs into belief function posteriors, effectively capturing epistemic uncertainty and offering an efficient and general methodology for uncertainty quantification. Comprehensive experiments employing a Bayesian Neural Network (BNN) baseline and an Interval Neural Network for inference on the MNIST, Fashion-MNIST, CIFAR-10 and CIFAR-100 datasets demonstrate that our Epistemic Wrapper significantly enhances generalisation and uncertainty quantification.

Figures

Figures reproduced from arXiv: 2505.02277 by the authors.

Figure 1
Figure 1. Epistemic Wrapper transforms weights posteriors from a Bayesian Neural Network into belief posteriors [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Posterior distribution of weights for the last layer. The histogram displays the sampled weights, overlaid with [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Graphical visualisation of the continuous PDF/mass function on intervals, with the area whose integral [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Probability densities of the Dirichlet distribution as functions on the 2D-simplex: [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Left: Scatter plot showing intervals projected onto a 2D simplex. Each point represents an interval A = [a, b] with its location determined by the values a and b, and the colour scale indicates the corresponding mass values m(A), ranging from 0.00 to 0.012. Right: Visu…
Figure 6
Figure 6. Figure 6: Accuracy Rejection Curve (ARC) for the MNIST dataset, comparing INN and Epi-Wrapper. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Varying Number of closed intervals layers employ Bayesian weight posterior distributions, where the kernel weights follow a Gaussian posterior parameterized by mean and variance. These distributions are constrained using a log-variance regularization technique, ensurin…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Epistemic Artificial Intelligence is Essential for Machine Learning Models to Truly 'Know When They Do Not Know'

    cs.AI 2025-05 conditional novelty 3.0 of 10

    Machine learning should use second-order uncertainty measures, such as credal sets and random sets, so models can explicitly represent ignorance and avoid overconfident predictions on unfamiliar data.

Reference graph

Works this paper leans on

54 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

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

    Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U Rajendra Acharya, et al. A review of uncertainty quantification in deep learning: Techniques, applications and challenges. Information Fusion, 76:243–297, 2021

  2. [2]

    Variational inference: A review for statisticians

    David M Blei, Alp Kucukelbir, and Jon D McAuliffe. Variational inference: A review for statisticians. Journal of the American statistical Association, 112(518):859–877, 2017

  3. [3]

    Weight uncertainty in neural network

    Charles Blundell, Julien Cornebise, Koray Kavukcuoglu, and Daan Wierstra. Weight uncertainty in neural network. In Proceedings of the International Conference on Machine Learning, pages 1613–1622. PMLR, 2015

  4. [4]

    Credal learning theory

    Michele Caprio, Maryam Sultana, Eleni Elia, and Fabio Cuzzolin. Credal learning theory. arXiv preprint arXiv:2402.00957, 2024

  5. [5]

    Molina, and Christopher Metzler

    Matthew Albert Chan, Maria J. Molina, and Christopher Metzler. Estimating epistemic and aleatoric uncertainty with a single model. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

  6. [6]

    Posterior network: Uncertainty estimation without ood samples via density-based pseudo-counts

    Bertrand Charpentier, Daniel Zügner, and Stephan Günnemann. Posterior network: Uncertainty estimation without ood samples via density-based pseudo-counts. Advances in neural information processing systems, 33:1356–1367, 2020

  7. [7]

    On the credal structure of consistent probabilities

    Fabio Cuzzolin. On the credal structure of consistent probabilities. In European Workshop on Logics in Artificial Intelligence, pages 126–139. Springer, 2008

  8. [8]

    Generalised max entropy classifiers

    Fabio Cuzzolin. Generalised max entropy classifiers. In Sébastien Destercke, Thierry Denœux, Fabio Cuzzolin, and Arnaud Martin, editors, Belief Functions: Theory and Applications , pages 39–47, Cham, 2018. Springer International Publishing

Show all 54 references
  1. [9]

    The geometry of uncertainty: The geometry of imprecise probabilities

    Fabio Cuzzolin. The geometry of uncertainty: The geometry of imprecise probabilities. Springer Nature, 2020

  2. [10]

    Uncertainty measures: The big picture

    Fabio Cuzzolin. Uncertainty measures: The big picture. arXiv preprint arXiv:2104.06839, 2021

  3. [11]

    Uncertainty measures: A critical survey

    Fabio Cuzzolin. Uncertainty measures: A critical survey. Information Fusion, page 102609, 2024

  4. [12]

    Belief modeling regression for pose estimation

    Fabio Cuzzolin and Wenjua Gong. Belief modeling regression for pose estimation. In Proceedings of the 16th International Conference on Information Fusion, pages 1398–1405, 2013

  5. [13]

    Upper and lower probabilities induced by a multivalued mapping

    Arthur P Dempster. Upper and lower probabilities induced by a multivalued mapping. In Classic works of the Dempster-Shafer theory of belief functions, pages 57–72. Springer, 2008

  6. [14]

    Large scale structure of neural network loss landscapes

    Stanislav Fort and Stanislaw Jastrzebski. Large scale structure of neural network loss landscapes. Advances in Neural Information Processing Systems, 32, 2019

  7. [15]

    Energy-based epistemic uncertainty for graph neural networks

    Dominik Fuchsgruber, Tom Wollschläger, and Stephan Günnemann. Energy-based epistemic uncertainty for graph neural networks. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. 11 A PREPRINT - AUGUST 20, 2025

  8. [16]

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

    Yarin Gal and Zoubin Ghahramani. Dropout as a Bayesian approximation: Representing model uncertainty in deep learning. In Proceedings of the International Conference on Machine Learning, pages 1050–1059. PMLR, 2016

  9. [17]

    A belief-theoretical approach to example-based pose estimation

    Wenjuan Gong and Fabio Cuzzolin. A belief-theoretical approach to example-based pose estimation. IEEE Transactions on Fuzzy Systems, 26(2):598–611, 2017

  10. [18]

    Interval arithmetic: From principles to implementation

    Timothy Hickey, Qun Ju, and Maarten H Van Emden. Interval arithmetic: From principles to implementation. Journal of the ACM (JACM), 48(5):1038–1068, 2001

  11. [19]

    The No-U-Turn sampler: Adaptively setting path lengths in Hamiltonian Monte Carlo

    Matthew D Hoffman, Andrew Gelman, et al. The No-U-Turn sampler: Adaptively setting path lengths in Hamiltonian Monte Carlo. J. Mach. Learn. Res., 15(1):1593–1623, 2014

  12. [20]

    Quantifying aleatoric and epistemic uncertainty: A credal approach

    Paul Hofman, Yusuf Sale, and Eyke Hüllermeier. Quantifying aleatoric and epistemic uncertainty: A credal approach. In ICML 2024 Workshop on Structured Probabilistic Inference{\&} Generative Modeling, 2024

  13. [21]

    J. R. M. Hosking. L-moments: Analysis and estimation of distributions using linear combinations of order statistics. Journal of the Royal Statistical Society: Series B (Methodological), 52(1):105–124, 12 2018

  14. [22]

    Aleatoric and epistemic uncertainty in machine learning: An introduc- tion to concepts and methods

    Eyke Hüllermeier and Willem Waegeman. Aleatoric and epistemic uncertainty in machine learning: An introduc- tion to concepts and methods. Machine Learning, 110(3):457–506, 2021

  15. [23]

    Hands-on Bayesian neural networks—A tutorial for deep learning users

    Laurent Valentin Jospin, Hamid Laga, Farid Boussaid, Wray Buntine, and Mohammed Bennamoun. Hands-on Bayesian neural networks—A tutorial for deep learning users. IEEE Computational Intelligence Magazine , 17(2):29–48, 2022

  16. [24]

    Is epistemic uncertainty faithfully represented by evidential deep learning methods? In Forty-first International Conference on Machine Learning, 2024

    Mira Juergens, Nis Meinert, Viktor Bengs, Eyke Hüllermeier, and Willem Waegeman. Is epistemic uncertainty faithfully represented by evidential deep learning methods? In Forty-first International Conference on Machine Learning, 2024

  17. [25]

    Understanding the difficulty of training deep feedforward neural networks xavier

    LS Kim. Understanding the difficulty of training deep feedforward neural networks xavier. In Proceedings of the International Joint Conference on Neural Networks, volume 2, 1993

  18. [26]

    Auto-encoding variational Bayes

    Diederik P Kingma and Max Welling. Auto-encoding variational Bayes. arXiv preprint arXiv:1312.6114, 2013

  19. [27]

    CIFAR-10 (Canadian Institute For Advanced Research)

    Alex Krizhevsky, Vinod Nair, and Geoffrey Hinton. CIFAR-10 (Canadian Institute For Advanced Research). 2009

  20. [28]

    Reliable confidence measures for medical diagnosis with evolutionary algorithms

    Antonis Lambrou, Harris Papadopoulos, and Alex Gammerman. Reliable confidence measures for medical diagnosis with evolutionary algorithms. IEEE Transactions on Information Technology in Biomedicine, 15(1):93– 99, 2010

  21. [29]

    The MNIST database of handwritten digits

    Yann LeCun. The MNIST database of handwritten digits. http: // yann. lecun. com/ exdb/ mnist/, 1998

  22. [30]

    The enterprise of knowledge: An essay on knowledge, credal probability, and chance

    Isaac Levi. The enterprise of knowledge: An essay on knowledge, credal probability, and chance. MIT press, 1980

  23. [31]

    Graph neural stochastic diffusion for estimating uncertainty in node classification

    Xixun Lin, Wenxiao Zhang, Fengzhao Shi, Chuan Zhou, Lixin Zou, Xiangyu Zhao, Dawei Yin, Shirui Pan, and Yanan Cao. Graph neural stochastic diffusion for estimating uncertainty in node classification. In Forty-first International Conference on Machine Learning, 2024

  24. [32]

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

    Jeremiah Liu, Zi Lin, Shreyas Padhy, Dustin Tran, Tania Bedrax Weiss, and Balaji Lakshminarayanan. Simple and principled uncertainty estimation with deterministic deep learning via distance awareness. Advances in Neural Information Processing Systems, 33:7498–7512, 2020

  25. [33]

    Evidence combination based on credal belief redistribution for pattern classification

    Zhun-Ga Liu, Yu Liu, Jean Dezert, and Fabio Cuzzolin. Evidence combination based on credal belief redistribution for pattern classification. IEEE Transactions on Fuzzy Systems, 28(4):618–631, 2019

  26. [34]

    Ensemble distribution distillation

    Andrey Malinin, Bruno Mlodozeniec, and Mark Gales. Ensemble distribution distillation. arXiv preprint arXiv:1905.00076, 2019

  27. [35]

    Epistemic deep learning

    Shireen Kudukkil Manchingal and Fabio Cuzzolin. Epistemic deep learning. arXiv preprint arXiv:2206.07609, 2022

  28. [36]

    A unified evaluation framework for epistemic predictions, 2025

    Shireen Kudukkil Manchingal, Muhammad Mubashar, Kaizheng Wang, and Fabio Cuzzolin. A unified evaluation framework for epistemic predictions, 2025

  29. [37]

    Random-set convolutional neural network (rs-cnn) for epistemic deep learning

    Shireen Kudukkil Manchingal, Muhammad Mubashar, Kaizheng Wang, Keivan Shariatmadar, and Fabio Cuzzolin. Random-set convolutional neural network (rs-cnn) for epistemic deep learning. arXiv preprint arXiv:2307.05772, 2023

  30. [38]

    Random-set neural networks

    Shireen Kudukkil Manchingal, Muhammad Mubashar, Kaizheng Wang, Keivan Shariatmadar, and Fabio Cuzzolin. Random-set neural networks. In The Thirteenth International Conference on Learning Representations, 2025. 12 A PREPRINT - AUGUST 20, 2025

  31. [39]

    Variational dropout sparsifies deep neural networks

    Dmitry Molchanov, Arsenii Ashukha, and Dmitry Vetrov. Variational dropout sparsifies deep neural networks. In Proceedings of the International Conference on Machine Learning, pages 2498–2507. PMLR, 2017

  32. [40]

    Adaptive sampling to reduce epistemic uncertainty using prediction interval-generation neural networks

    Giorgio Morales and John Sheppard. Adaptive sampling to reduce epistemic uncertainty using prediction interval-generation neural networks. Proceedings of the AAAI Conference on Artificial Intelligence, Feb. 2025

  33. [41]

    MCMC using Hamiltonian dynamics

    Radford M Neal et al. MCMC using Hamiltonian dynamics. Handbook of Markov Chain Monte Carlo, 2(11):2, 2011

  34. [42]

    Turner, Rio Yokota, and Mohammad Emtiyaz Khan

    Kazuki Osawa, Siddharth Swaroop, Anirudh Jain, Runa Eschenhagen, Richard E. Turner, Rio Yokota, and Mohammad Emtiyaz Khan. Practical deep learning with bayesian principles. arXiv preprint arXiv:1906.02506, 2019

  35. [43]

    Is the volume of a credal set a good measure for epistemic uncertainty? In Uncertainty in Artificial Intelligence, pages 1795–1804

    Yusuf Sale, Michele Caprio, and Eyke Höllermeier. Is the volume of a credal set a good measure for epistemic uncertainty? In Uncertainty in Artificial Intelligence, pages 1795–1804. PMLR, 2023

  36. [44]

    Evidential deep learning to quantify classification uncertainty

    Murat Sensoy, Lance Kaplan, and Melih Kandemir. Evidential deep learning to quantify classification uncertainty. Advances in neural information processing systems, 31, 2018

  37. [46]

    A mathematical theory of evidence, volume 42

    Glenn Shafer. A mathematical theory of evidence, volume 42. Princeton university press, 1976

  38. [47]

    The transferable belief model and other interpretations of Dempster–Shafer’s model

    Philippe Smets. The transferable belief model and other interpretations of Dempster–Shafer’s model. In P. P. Bonissone, M. Henrion, L. N. Kanal, and J. F. Lemmer, editors,Uncertainty in Artificial Intelligence, volume 6, pages 375–383. North-Holland, Amsterdam, 1991

  39. [48]

    Belief functions on real numbers.International Journal of Approximate Reasoning, 40(3):181–223, 2005

    Philippe Smets. Belief functions on real numbers.International Journal of Approximate Reasoning, 40(3):181–223, 2005

  40. [49]

    Thiagarajan

    Puja Trivedi, Mark Heimann, Rushil Anirudh, Danai Koutra, and Jayaraman J. Thiagarajan. Accurate and scalable estimation of epistemic uncertainty for graph neural networks. In The Twelfth International Conference on Learning Representations, 2024

  41. [50]

    Credal wrapper of model averaging for uncertainty estimation on out-of-distribution detection

    Kaizheng Wang, Fabio Cuzzolin, Keivan Shariatmadar, David Moens, and Hans Hallez. Credal wrapper of model averaging for uncertainty estimation on out-of-distribution detection. arXiv preprint arXiv:2405.15047, 2024

  42. [51]

    Credal deep ensembles for uncertainty quantification

    Kaizheng Wang, Fabio Cuzzolin, Keivan Shariatmadar, David Moens, Hans Hallez, et al. Credal deep ensembles for uncertainty quantification. Advances in Neural Information Processing Systems, 37:79540–79572, 2024

  43. [52]

    Creinns: Credal-set interval neural networks for uncertainty estimation in classification tasks

    Kaizheng Wang, Keivan Shariatmadar, Shireen Kudukkil Manchingal, Fabio Cuzzolin, David Moens, and Hans Hallez. Creinns: Credal-set interval neural networks for uncertainty estimation in classification tasks. Neural Networks, page 107198, 2025

  44. [53]

    Wasserman

    Larry A. Wasserman. Belief functions and statistical inference. Canadian Journal of Statistics, 18(3):183–196, 1990

  45. [54]

    Flipout: Efficient pseudo-independent weight perturbations on mini-batches

    Yeming Wen, Paul Vicol, Jimmy Ba, Dustin Tran, and Roger Grosse. Flipout: Efficient pseudo-independent weight perturbations on mini-batches. In Proceedings of the International Conference on Learning Representations , 2018

  46. [55]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747, 2017. 13 A PREPRINT - AUGUST 20, 2025 A Appendix A.1 Interval Neural Networks (INNs) Traditional interval neural ne...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.