Pith. sign in

REVIEW 2 major objections 6 minor 24 references

Uncertainty quality tracks multi-modal diversity: methods that explore many loss basins set the standard, while shared-weight shortcuts and single-pass tricks trade that diversity away.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-31 13:22 UTC pith:GUBLO43X

load-bearing objection Solid scoped survey: method–measure split and mechanism×scope taxonomy are the real value; L/M/H tables are disclosed qualitative placements, not a hidden flaw. the 2 major comments →

arxiv 2607.28248 v1 pith:GUBLO43X submitted 2026-07-30 stat.ML cs.LG

Uncertainty quantification for trustworthy deep learning: Methods and measures

H. Martin Gillis , Thomas Trappenberg This is my paper
classification stat.ML cs.LG
keywords uncertainty quantificationepistemic uncertaintyensemble diversitycalibrationout-of-distribution detectiondeep ensemblessingle-pass uncertaintylarge language models
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Deep networks used in safety-critical settings need honest confidence scores, not just point predictions. This survey argues that the field is clearer once you split two stages that papers usually mix: the method that produces a set of predictions, and the measure that turns that set into total, aleatoric, or epistemic uncertainty. Along a simple grid of how the ensemble is made (explicit members, posterior samples, or a closed-form single pass) and how much of the network is free (full net, last layer, or one deterministic model), the pattern is consistent. Full-network methods that land in different loss basins—deep ensembles and their multi-modal cousins—remain the quality benchmark for calibration and out-of-distribution detection. Efficient approximations that share weights, stay in one basin, or read uncertainty from a single forward pass cut cost but usually cut functional diversity, and with it shift robustness; distance-aware single-pass designs can still score well on OOD without true between-mode diversity. The same split exposes open gaps: better last-layer diversity, cheap epistemic scores for classification that do not rely on fragile entropy identities, hybrids that add a little multi-modal diversity to strong single-pass backbones, and principled uncertainty for free-form language generation.

Core claim

Reading uncertainty quantification through a method–measure split and a generative-mechanism × network-scope taxonomy shows that between-mode, full-network exploration is what delivers top-tier uncertainty, while efficient shared-weight and single-basin methods systematically lose functional diversity and therefore weaker behavior under shift—yet carefully designed distance-aware single-pass models can still reach strong OOD detection without that multi-modal component.

What carries the argument

The method–measure separation plus the two-axis taxonomy (how the predictive ensemble is generated: explicit members, posterior sampling, or closed-form predictive; and network scope: full network, shared backbone/last layer, or single deterministic net). That grid, not architecture brand names, is what organizes the quality–cost trade-off and the comparison tables.

Load-bearing premise

That qualitative high/medium/low ratings stitched from different papers, architectures, datasets, and shifts—including some inferred only from method structure—are solid enough to rank whole method families head to head.

What would settle it

On a fixed architecture and the standard OpenOOD / shift suites, show that a cheap shared-weight or last-layer method matches a deep ensemble on both calibration under corruption severity and OOD AUROC once compute is matched; or show that last-layer diversity alone recovers most of the ensemble’s epistemic gap.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Practitioners should treat deep ensembles (and multi-modal extensions) as the quality reference and judge cheaper methods by how much functional diversity they keep under shift.
  • Any sample-based ensemble can be paired with any sample-based uncertainty measure; closed-form single-pass methods only expose their built-in signal unless you sample them.
  • Hybrid designs—small ensembles of distance-aware backbones, or one distance-aware trunk with several heads—become the natural next architecture class to test.
  • Bounded pairwise scores and linear-time gated epistemic measures need the same standardized OOD/shift bake-off that methods already get.
  • Generative language uncertainty must redefine diversity and the aleatoric/epistemic split over meanings, not fixed labels.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If diversity is the real scarce resource, evaluation suites should report effective ensemble size or representation disagreement under shift, not only ECE and AUROC.
  • Safety cases that rely on single-pass deterministic uncertainty may be fine for far-OOD flagging yet still miss the multi-solution ignorance that full ensembles capture.
  • The method–measure split suggests a modular software stack: one backend that emits member predictives, and swappable measure plugins—including conformal wrappers—without retraining.
  • Epistemic collapse at large scale implies that “more parameters” alone will not buy trustworthy uncertainty; explicit multi-modal or repulsive mechanisms may stay necessary.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. This survey reviews uncertainty quantification for deep learning, scoped to ensemble-based and approximate Bayesian methods and to the measures that summarize their predictive ensembles. Its organizing contribution is a clean separation between the method that produces a predictive distribution and the measure that reads uncertainty from it, together with a two-axis taxonomy (generative mechanism × network scope; Table 2, Figure 1). Methods are grouped into BNNs, MC Dropout, deep ensembles, efficient ensemble approximations, and last-layer/single-pass approaches, with adjacent treatment of evidential/prior networks, conformal prediction, post-hoc calibration, OOD detection, selective prediction, and a brief LLM section. The authors consolidate evaluation practice (Section 11), diversity theory (Section 13), and entropy versus pairwise-divergence measures (Section 14), and close with open directions on last-layer diversity, efficient epistemic measures, shift, hybrids, and generative UQ.

Significance. If the organizational claims hold—as the loss-landscape and diversity literature largely supports—the paper offers a usable map of a crowded field that existing surveys treat more broadly or more architecture-first. Depth on efficient ensembles (BatchEnsemble, SWAG/MultiSWAG, CreDE, masksembles) and single-pass/last-layer methods (SNGP, DUQ/DDU/DUE, LLE, VBLL), plus an explicit method–measure decoupling and a consolidated evaluation basis for qualitative comparisons, are genuine contributions relative to Gawlikowski et al., Abdar et al., and He et al. The structural reading that full-network multi-modal exploration sets the quality standard while shared-weight/single-basin approximations lose functional diversity, with distance-aware single-pass methods as a partial OOD exception, is a clear and citable synthesis. Open problems (efficient non-entropy epistemic measures, last-layer diversity limits, hybrids, LLM semantic UQ) are well posed for follow-on work.

major comments (2)
  1. [Section 11–12, Table 3] Section 11–12 and Table 3 note (a)/†: the L/M/H OOD and calibration ratings are the main comparative device supporting family rankings, yet they pool heterogeneous architectures, datasets, and shift types, with several † ratings inferred from structure rather than direct benchmarks. The manuscript already discloses this and frames the cells as relative placements, which is appropriate for a survey; still, the gold-standard narrative and several M/H contrasts would be more robust if Table 3 (or a short appendix) marked which cells rest on OpenOOD/Uncertainty Baselines-style head-to-heads versus originating-paper or structural inference, so readers can weight the rankings accordingly.
  2. [Section 14.5–14.7, Tables 6–8] Section 14.5–14.7 and Tables 6–8: the structural case that EPJS removes EPKL’s unboundedness failure mode while avoiding MI’s mixture underestimation is carefully stated as non-empirical, and validation on standard OOD/shift suites is correctly listed as open (Section 17.2). Because the survey’s measure contribution partly rests on elevating pairwise and variance-gated alternatives over the default MI decomposition, a brief explicit caveat in Section 14.5 (and in Table 7’s EPJS/VGMU rows) that no claim of superior detection AUROC is being made would prevent over-reading by practitioners.
minor comments (6)
  1. [Section 14.7] Section 14.7 and Table 5/7: variance-gated ensembles, VGMU, and VGN are introduced largely via Gillis et al. (2025, 2026b). A one-sentence note that these are recent proposals with limited independent benchmarking (parallel to the CreDE note in Table 3) would match the paper’s care elsewhere and reduce any appearance of uneven scrutiny.
  2. [Figure 1] Figure 1 is dense; the quality color scale and the closed-form span across scopes are useful but hard to parse in grayscale. A simplified schematic in the main text and a fuller version in the supplement (or clearer legend encoding) would help.
  3. [Table 1] Table 1 timeline is helpful; a few adjacent items (e.g., conformalized quantile regression, temperature scaling) sit under “Evidential, conformal, and calibration” while core families are year-grouped—consider a single chronological column or clearer subgroup headers for skimming.
  4. [Section 16] Section 16 is appropriately scoped as adjacent, but the jump from classification ensembles to semantic entropy could cite Malinin & Gales (2021) earlier when first mentioning autoregressive decomposition, to link back to Section 14.
  5. [Section 3, Eq. (7)] Minor notation: p(w|D) vs p(w| D) spacing is inconsistent in Section 3; EPCE uses H[p,q] for cross-entropy while H also denotes entropy—define cross-entropy explicitly at first use in Eq. (7).
  6. [Section 1.1, 5.1, title block] Typos/style: “amodeling convention” (Section 1.1) needs a space; “thede factogold standard” (Section 5.1) needs spaces; author email “martin.gillis@.dal.ca” appears to drop a character.

Circularity Check

1 steps flagged

Survey organization is not circular; only mild non-load-bearing self-citation of the authors’ variance-gated and last-layer work.

specific steps
  1. self citation load bearing [Section 14.7; also Section 7.4 and Table 5/7 (Gillis et al. 2025, 2026a, 2026b)]
    "To reduce this cost, Variance-Gated Ensembles (VGEs) (Gillis et al., 2025, 2026b) gate member probabilities by a signal-to-noise factor derived from the ensemble mean and per-class predictive spread, yielding a total/aleatoric/epistemic decomposition at O(MC), a margin-based epistemic score, Variance-Gated Margin Uncertainty (VGMU)..."

    Authors cite their own concurrent/recent work when placing VGE/VGMU/VGN (and last-layer committee machines) in the measure/method landscape. This is ordinary survey self-reference, not a load-bearing uniqueness or derivation step: the survey’s main claims (deep ensembles as multi-modal gold standard; method–measure decoupling; diversity–quality link) do not reduce to these citations and remain supported by independent external sources.

full rationale

This is a scoped survey whose central contribution is taxonomic and comparative (method vs. measure separation; generative-mechanism × network-scope placement in Table 2/Figure 1; qualitative L/M/H ratings anchored to external benchmarks in Section 11). It does not claim a first-principles derivation, uniqueness theorem, or fitted-parameter “prediction.” Comparative claims rest on external literature (e.g., Fort et al. 2020; Ovadia et al. 2019; Wood et al. 2023; Zamyatin et al. 2026) and on structural placement of methods, not on closing a definitional loop. Self-citations to Gillis et al. (2025, 2026a, 2026b) appear only when situating variance-gated measures and last-layer committee machines inside the existing landscape; they are not used to force the gold-standard narrative or the taxonomy. Table 3’s heterogeneous/† ratings are a transparency and evidentiary-strength issue, not circularity by construction. No equation reduces a claimed prediction to its own fitted input. Score 1 reflects only routine author self-reference that is not load-bearing.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

As a survey, load-bearing commitments are definitional and evaluative conventions from the UQ literature rather than fitted physical constants. The central narrative depends on treating aleatoric/epistemic splits as model-relative, equating high-quality epistemic UQ with functional/multi-modal diversity, and accepting heterogeneous published benchmarks as commensurable enough for L/M/H family ratings.

axioms (5)
  • domain assumption Predictive uncertainty decomposes into aleatoric and epistemic components via entropy/MI or variance laws, understood as a modeling convention rather than an intrinsic data property (Sections 1.1, 14).
    Organizes the entire measure discussion; paper correctly notes the split is convention-dependent.
  • domain assumption Between-mode functional diversity from full-network explicit members is the primary driver of strong epistemic UQ under distribution shift (Sections 5–6, 12–13).
    Underpins gold-standard status of deep ensembles and critiques of shared-weight approximations.
  • ad hoc to paper Qualitative OOD/calibration tiers can be anchored to MSP vs deep-ensemble performance on standard pairs and Ovadia-style shift suites across heterogeneous studies (Section 11, Table 3).
    Enables head-to-head tables without a unified re-benchmark; paper discloses limits and † inferences.
  • domain assumption Sample-based measures are freely interchangeable across methods that expose member/posterior samples; closed-form single-pass methods expose only intrinsic signals unless sampled (Figure 1, Section 17).
    Core method–measure separation claim of the survey.
  • standard math Standard information-theoretic and ensemble-diversity identities (predictive entropy, MI, bias–variance–diversity, pairwise KL/JS) hold as used in the cited sources.
    Background mathematics for Sections 13–14.

pith-pipeline@v1.2.0-daily-grok45 · 42968 in / 3203 out tokens · 59395 ms · 2026-07-31T13:22:16.578067+00:00 · methodology

0 comments
read the original abstract

The deployment of deep neural networks in safety-critical domains demands reliable estimates of predictive confidence, yet conventional architectures lack principled uncertainty quantification. This survey provides a structured, critical review of methods for Uncertainty Quantification (UQ) in deep learning, scoped to ensemble-based and approximate Bayesian approaches and the measures used to summarize their outputs. Relative to existing UQ surveys, our contribution is depth on efficient ensemble approximations and single-pass methods, and a unified treatment that separates the method producing a predictive distribution from the measure that summarizes its uncertainty. We organize methods into five families: Bayesian neural networks, Monte Carlo Dropout, deep ensembles, efficient ensemble approximations, and last-layer or single-pass approaches. We situate adjacent work on evidential and prior networks, conformal prediction, and post-hoc calibration, together with the decision-time tasks of out-of-distribution detection and selective prediction. For each, we examine theoretical motivation, implementation, empirical performance, and limitations. We then review ensemble diversity theory and uncertainty measures and their decompositions, contrasting the entropy decomposition with pairwise divergence measures, and consolidate evaluation methodology so that our qualitative comparisons share a common basis. We close with a brief treatment of uncertainty in large language models and open research directions, including efficient epistemic measures for classification, last-layer diversity, diversity and calibration under shift, and hybrid architectures.

Figures

Figures reproduced from arXiv: 2607.28248 by H. Martin Gillis, Thomas Trappenberg.

Figure 1
Figure 1. Figure 1: Overview of uncertainty quantification (UQ) in deep learning. Panel (a). Taxonomy used throughout the survey. UQ methods (Section 3–Section 12) produce a predictive ensemble; uncertainty measures (Section 14) summarize that ensemble into total, aleatoric, epistemic, and efficient quantities, and support decision-time tasks such as OOD detection, selective prediction, and active learning (Section 15). Metho… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

24 extracted references · 3 canonical work pages · 1 internal anchor

  1. [7]

    Workshop: Bayesian deep learning

    URL:https://bayesiandeeplearning.org/2019/ papers/39.pdf. Workshop: Bayesian deep learning. Valdenegro-Toro, M., 2023. Sub-ensembles for fast uncertainty estimation in neural networks, in: International Conference on Computer Vision Workshops (ICCVW), pp. 4119–4127. doi:10.1109/ICCVW60793.2023.00445. Valiuddin, A., van Sloun, R., Viviers, C., de With, P.,...

    arXiv 2019

  2. [11]

    Monod, M., Micheli, A., Bhatt, S., 2025

    doi:10.1038/s41598-021-84854-x. Monod, M., Micheli, A., Bhatt, S., 2025. NeuralSurv: Deep survival analysis with Bayesian uncertainty quantification, in: Conference on Neural Information Processing Systems (NeurIPS), pp. 1–53. URL:https://openreview.net/ pdf?id=c768Z1FwDL. Mukhoti, J., Kirsch, A., van Amersfoort, J., Torr, P.H.S., Gal, Y ., 2023. Deep det...

    arXiv 2025

  3. [12]

    Workshop: Bayesian Deep Learning

    URL:https://bayesiandeeplearning.org/2021/ papers/21.pdf. Workshop: Bayesian Deep Learning. Naeini, M.P., Cooper, G.F., Hauskrecht, M., 2015. Obtain- ing well-calibrated probabilities using Bayesian binning, in: AAAI Conference on Artificial Intelligence, pp. 2901–2907. doi:10.1609/aaai.v29i1.9602. Nandy, J., Hsu, W., Lee, M.L., 2021. Towards maximizing t...

    arXiv 2021

  4. [16]

    Workshop: Bayesian deep learning

    URL:https://bayesiandeeplearning.org/2021/ papers/28.pdf. Workshop: Bayesian deep learning. van Amersfoort, J., Smith, L., Teh, Y .W., Gal, Y ., 2020. Uncertainty estimation using a single deep deter- ministic neural network, in: International Confer- ence on Machine Learning (ICML), pp. 9690–9700. URL:https://proceedings.mlr.press/v119/ van-amersfoort20a...

    Pith/arXiv arXiv doi:10.48550/arxiv.2309.16314 2021

  5. [18]

    ACM Computing Surveys 58, 1–38

    A survey on uncertainty quantification of large lan- guage models: Taxonomy, open research challenges, and fu- ture directions. ACM Computing Surveys 58, 1–38. doi:10. 1145/3744238. Smith, L., Gal, Y ., 2018. Understanding measures of uncer- tainty for adversarial example detection. arXiv preprint. doi:10.48550/arXiv.1803.08533. Srivastava, N., Hinton, G....

    Pith/arXiv arXiv doi:10.48550/arxiv.1803.08533 2018

  6. [19]

    Function space diversity for uncertainty prediction via repulsive last-layer ensembles, in: International Con- ference on Machine Learning (ICML), pp. 1–16. URL: https://openreview.net/pdf?id=FbMN9HjgHI. Work- shop: Structured Probabilistic Inference & Generative Mod- eling. Tagasovska, N., Lopez-Paz, D., 2019. Single-model uncertain- ties for deep learni...

    arXiv 2019

  7. [22]

    A rigorous link between deep ensembles and (vari- ational) Bayesian methods, in: Conference on Neural In- formation Processing Systems (NeurIPS), arXiv. pp. 1–30. URL:https://openreview.net/pdf?id=eTHawKFT4h& noteId=UBYXOwJMPl. Wilson, A.G., Izmailov, P., 2020. Bayesian deep learning and a probabilistic perspective of generalization, in: Conference on Neu...

    arXiv 2020

  8. [40]

    Rahaman, R., Thiery, A.H., 2021

    URL:https://proceedings.mlr.press/v162/ postels22a/postels22a.pdf. Rahaman, R., Thiery, A.H., 2021. Uncertainty quantification and deep ensembles, in: Conference on Neural Information Processing Systems (NeurIPS), pp. 1–13. URL:https:// openreview.net/pdf?id=wg_kD_nyAF. Ramé, A., Cord, M., 2021. DICE: Diversity in deep en- sembles via conditional redundan...

    2021

  9. [49]

    Xie, J., Ma, Z., Lei, J., Zhang, G., Xue, J.H., Tan, Z.H., Guo, J., 2022

    URL:https://www.jmlr.org/papers/volume24/ 23-0041/23-0041.pdf. Xie, J., Ma, Z., Lei, J., Zhang, G., Xue, J.H., Tan, Z.H., Guo, J., 2022. Advanced dropout: A model-free methodology for Bayesian dropout optimization. IEEE Transactions on Pattern Analysis and Machine Intelligence 44, 4605–4625. doi:10.1109/TPAMI.2021.3083089. Xiong, M., Hu, Z., Lu, X., Li, Y...

    arXiv 2022

  10. [1367]

    Chau, S.L., Caprio, M., Muandet, K., 2025

    URL:https://dl.acm.org/doi/pdf/10.5555/ 3495724.3495839. Chau, S.L., Caprio, M., Muandet, K., 2025. Integral imprecise probability metrics, in: Conference on Neural Information Processing Systems (NeurIPS), pp. 1–42. URL:https:// openreview.net/pdf?id=KM2XzHq2Rm. Chauhan, V .K., Zhou, J., Lu, P., Molaei, S., Clifton, D.A.,

    arXiv 2025

  11. [1622]

    Boluki, S., Ardywibowo, R., Dadaneh, S.Z., Zhou, M., Qian, X., 2020

    URL:https://proceedings.mlr.press/v37/ blundell15.pdf. Boluki, S., Ardywibowo, R., Dadaneh, S.Z., Zhou, M., Qian, X., 2020. Learnable bernoulli dropout for Bayesian deep learning, in: International Conference on Arti- ficial Intelligence and Statistics (AISTATS), pp. 3905–

    2020

  12. [1914]

    Farquhar, S., Kossen, J., Kuhn, L., Gal, Y ., 2024

    URL:https://ieeexplore.ieee.org/stamp/ stamp.jsp?arnumber=9956231. Farquhar, S., Kossen, J., Kuhn, L., Gal, Y ., 2024. De- tecting hallucinations in large language models using se- mantic entropy. Nature 630, 625–630. doi:10.1038/ s41586-024-07421-0. Felicioni, N., Maystre, L., Ghiassian, S., Ciosek, K., 2024. On the importance of uncertainty in decision-...

    Pith/arXiv arXiv doi:10.48550/arxiv.1912.02757 2024

  13. [2005]

    Active learning for probability estimation using Jensen-Shannon divergence, in: European Conference on Machine Learning, pp. 268–279. doi:10.1007/11564096_ 28. Mena, J., Pujol, O., Vitrià, J., 2021. A survey on uncertainty estimation in deep learning classification systems from a Bayesian perspective. ACM Computing Surveys 54, 1–35. doi:10.1145/3477140. 2...

    work page doi:10.1007/11564096_ 2021

  14. [2015]

    Weight uncertainty in neural networks, in: Interna- tional Conference on Machine Learning (ICML), pp. 1613–

  15. [2020]

    On Last-Layer Algorithms for Classification: Decoupling Representation from Uncertainty Estimation

    On last-layer algorithms for classification: Decoupling representation from uncertainty estimation. arXiv preprint. doi:10.48550/arXiv.2001.08049. Brown, G., Wyatt, J., Harris, R., Yao, X., 2005. Diversity cre- ation methods: A survey and categorisation. Information Fu- sion 6, 5–20. doi:10.1016/j.inffus.2004.04.004. Charpentier, B., Zügner, D., Günnemann...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2001.08049 2001

  16. [2023]

    Post-hoc uncertainty learning using a Dirichlet meta- model, in: Conference on Artificial Intelligence, pp. 9772–

  17. [2024]

    Artificial Intelligence Review 57

    A brief review of hypernetworks in deep learn- ing. Artificial Intelligence Review 57. doi:10.1007/ s10462-024-10862-8. Damianou, A.C., Lawrence, N.D., 2013. Deep gaussian process, in: International Conference on Artificial Intelli- gence and Statistics (AISTATS), pp. 1–9. URL:https: //proceedings.mlr.press/v31/damianou13a.pdf. Daxberger, E., Kristiadi, A...

    Pith/arXiv arXiv doi:10.48550/arxiv.1802.04865 2013

  18. [2061]

    Liang, J., Hou, R., Hu, M., Chang, H., Shan, S., Chen, X., 2025

    URL:https://proceedings.mlr.press/v70/ li17a/li17a.pdf. Liang, J., Hou, R., Hu, M., Chang, H., Shan, S., Chen, X., 2025. Revisiting logit distributions for reliable out-of- distribution detection, in: Conference on Neural Informa- tion Processing Systems (NeurIPS), pp. 1–33. URL:https: //openreview.net/pdf?id=FLdLPUqnsP. Liang, S., Li, Y ., Srikant, R., 2...

    arXiv 2025

  19. [2292]

    Wood, D., Mu, T., Webb, A.M., Reeve, H.W.J., Luján, M., Brown, G., 2023

    URL:https://proceedings.mlr.press/v216/ wimmer23a/wimmer23a.pdf. Wood, D., Mu, T., Webb, A.M., Reeve, H.W.J., Luján, M., Brown, G., 2023. A unified theory of diversity in ensem- ble learning. Journal of Machine Learning Research 24, 1– 26

    2023

  20. [3467]

    Valdenegro-Toro, M., 2019

    URL:https://proceedings.mlr.press/v89/ vaicenavicius19a/vaicenavicius19a.pdf. Valdenegro-Toro, M., 2019. Deep sub-ensembles for fast un- certainty estimation in image classification, in: Conference on Neural Information Processing Systems (NeurIPS), pp. 1–

    2019

  21. [3640]

    Workshop: Mathematics of modern machine learning

    URL:https://raw.githubusercontent.com/ mlresearch/v286/main/assets/schweighofer25a/ schweighofer25a.pdf. Workshop: Mathematics of modern machine learning. Schweighofer, K., Aichberger, L., Ielanskyi, M., Klambauer, G., Hochreiter, S., 2023b. Quantification of uncertainty with adversarial models, in: Neural Information Processing Sys- tems (NeurIPS), pp. 1...

    doi:10.48550/arxiv.2201 2022

  22. [3916]

    19 Brier, G.W., 1950

    URL:https://proceedings.mlr.press/v108/ boluki20a/boluki20a.pdf. 19 Brier, G.W., 1950. Verification of forecasts expressed in terms of probability. Monthly Weather Review 78, 1–3. doi:10. 1175/1520-0493(1950)078<0001:VOFEIT>2.0.CO;2. Brosse, N., Riquelme, C., Martin, A., Gelly, S., Moulines, E.,

    1950

  23. [7058]

    Malinin, A., Gales, M., 2019

    URL:https://dl.acm.org/doi/pdf/10.5555/ 3327757.3327808. Malinin, A., Gales, M., 2019. Reverse KL-divergence training of prior networks: Improved uncertainty and adversarial robustness, in: Conference on Neural Information Pro- cessing Systems (NeurIPS), pp. 1–12. URL:https:// proceedings.neurips.cc/paper_files/paper/2019/ file/7dd2ae7db7d18ee7c9425e38df1...

    arXiv 2019

  24. [9781]

    Shorinwa, O., Mei, Z., Lidard, J., Ren, A.Z., Majumdar, A.,

    doi:10.1609/aaai.v37i8.26167. Shorinwa, O., Mei, Z., Lidard, J., Ren, A.Z., Majumdar, A.,

    work page doi:10.1609/aaai.v37i8.26167