pith. sign in

arxiv: 2507.04441 · v4 · submitted 2025-07-06 · 📊 stat.ML · cs.AI· cs.LG· math.CT

A Category-Theoretic Analysis of Conformal Prediction

Michele Caprio This is my paper

Pith reviewed 2026-05-19 06:02 UTC · model grok-4.3

classification 📊 stat.ML cs.AIcs.LGmath.CT
keywords conformal predictioncategory theorypredictive distributionsBayesian compatibilityuncertainty quantificationasymptotic convergenceprivacy preservationset-valued procedures
0
0 comments X

The pith

Full conformal prediction corresponds to morphisms in categories of stable set-valued procedures and measurable random regions, decomposed by a commuting diagram into distribution extraction followed by region derivation.

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

The paper develops a category-theoretic approach to make the structure of conformal prediction explicit. Full conformal prediction is represented as a morphism in two categories that capture stability of set-valued procedures and measurability of random regions. A commuting diagram result decomposes the construction into extracting predictive distributions from the data and then deriving a prediction region from that set. This decomposition provides a route to numerical uncertainty summaries beyond region size. The work also establishes asymptotic compatibility showing that conformal regions converge to Bayesian predictive density level sets under local empirical process and boundary regularity assumptions, with quantitative rates.

Core claim

Full Conformal Prediction can be represented as a morphism in two categories capturing stability of set-valued procedures and measurability of random regions. Under mild conditions a commuting diagram decomposes the construction into extracting a set of predictive distributions from the data and then deriving a prediction region from this set. This yields asymptotic compatibility to Bayesian predictive density level sets under local empirical process and boundary regularity assumptions, with quantifiable convergence rates, while the region extractor is shown to be functorial.

What carries the argument

The commuting diagram in the category-theoretic representation of full conformal prediction that separates the extraction of predictive distributions from the subsequent derivation of the prediction region.

If this is right

  • The decomposition supplies a principled route to numerical uncertainty summaries beyond the size of the prediction region.
  • Conformal regions converge to Bayesian predictive density level sets with quantitative rates under the local empirical process and boundary regularity assumptions.
  • Upper posterior constructions relate to e-posteriors under identified conditions, clarifying when e-value-based and conformal-imprecise representations coincide.
  • Functoriality of the region extractor supports a modular privacy-compatible perspective in which privacy-preserving outer approximations of shared summary objects produce conservative global prediction regions.

Where Pith is reading between the lines

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

  • The categorical decomposition could allow modular composition of conformal procedures with other statistical functors representing different forms of uncertainty quantification.
  • Viewing conformal sets categorically may strengthen links to imprecise probability by treating prediction regions as objects in a category of set-valued maps.
  • The framework suggests testable extensions such as applying the same commuting diagram analysis to bootstrap or jackknife-based prediction methods.

Load-bearing premise

The data-generating process satisfies local empirical process conditions and boundary regularity so that conformal regions converge to Bayesian level sets at quantifiable rates.

What would settle it

Simulate data from a distribution that violates boundary regularity and check whether the conformal prediction regions fail to approach the corresponding Bayesian predictive density level sets at the rates claimed.

Figures

Figures reproduced from arXiv: 2507.04441 by Michele Caprio.

Figure 1
Figure 1. Figure 1: A visual representation of Full CP methodology (Caprio et al., 2025a, [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A visual representation of the two-step procedure that first extracts credal set M(Π) from the training set y n as in (7), and then computes the IHDR as in (6) (Caprio et al., 2025a, [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
read the original abstract

Conformal prediction (CP) produces prediction regions with finite-sample, distribution free coverage guarantees, but its interpretation as a quantitative uncertainty tool is often left implicit. We develop a category-theoretic approach that makes this structure explicit. We show that Full Conformal Prediction can be represented as a morphism in two categories capturing (i) stability of set-valued procedures and (ii) measurability of random regions. Under mild conditions, we prove a commuting diagram result that decomposes the construction of a conformal region into two steps: Extracting a set of predictive distributions from the data, and then deriving a prediction region from this set. This decomposition provides a principled route to numerical uncertainty summaries beyond region size. We further prove an asymptotic compatibility result showing that, for Bayesian predictive scores in regular regimes, conformal regions converge to Bayesian predictive density level sets; We also provide quantitative rates under local empirical process and boundary regularity assumptions. This highlights a bridge between Bayesian, frequentist, and imprecise probabilistic prediction. We additionally identify conditions under which upper posterior constructions are related to e-posteriors, clarifying when e-value-based and conformal-imprecise representations can coincide. Finally, we show that the region extractor is functorial; This yields a modular privacy-compatible perspective in which privacy-preserving outer approximations of shared summary objects lead to conservative global prediction regions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper develops a category-theoretic framework for Full Conformal Prediction (CP). It represents CP as morphisms in two categories: one capturing stability of set-valued procedures and another for measurability of random regions. Under mild conditions, it proves a commuting diagram that decomposes conformal region construction into extracting a set of predictive distributions from data and then deriving a prediction region. It further establishes an asymptotic compatibility result showing that conformal regions converge to Bayesian predictive density level sets under local empirical process and boundary regularity assumptions, with quantitative rates. The paper also identifies conditions relating upper posterior constructions to e-posteriors and proves that the region extractor is functorial, yielding a modular privacy-compatible perspective via conservative outer approximations.

Significance. If the central claims hold, the work provides a novel bridge between conformal prediction, Bayesian methods, and imprecise probabilities through category theory, with the commuting diagram offering a principled decomposition for uncertainty summaries and the functoriality result enabling modular privacy-preserving inference. The quantitative asymptotic rates are a strength when the regularity conditions apply. These contributions could clarify interpretations of CP as a quantitative uncertainty tool and suggest new connections across paradigms, though their impact depends on the generality of the invoked assumptions.

major comments (2)
  1. [Theorem 5.1] Theorem 5.1 (asymptotic compatibility): the claimed quantitative rates and convergence to Bayesian level sets rest on local empirical process and boundary regularity assumptions, but the manuscript does not provide counterexamples, simulation evidence, or scope analysis showing these conditions hold beyond smooth unimodal cases (e.g., multimodal densities or non-Lipschitz boundaries); this is load-bearing for the bridge between paradigms asserted in the abstract and §5.
  2. [§4.2] §4.2 (commuting diagram): while the decomposition into predictive distribution extraction and region derivation is presented as functorial, the proof does not explicitly verify that the diagram commutes while preserving the finite-sample coverage guarantee of CP; an explicit natural transformation or worked example with a standard nonconformity score (e.g., absolute residual) is needed to confirm the construction is not merely formal.
minor comments (2)
  1. [§3] The definitions of the two categories (stability and measurability) appear in §3 but could benefit from an earlier, self-contained summary of objects and morphisms to aid readers unfamiliar with category theory in statistics.
  2. [Throughout] Notation for random regions and set-valued procedures is introduced without a consolidated table of symbols; adding one would improve readability across the commuting diagram and functoriality results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We appreciate the referee's insightful comments, which help improve the clarity and rigor of our category-theoretic framework for conformal prediction. We respond to each major comment below, indicating planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Theorem 5.1] Theorem 5.1 (asymptotic compatibility): the claimed quantitative rates and convergence to Bayesian level sets rest on local empirical process and boundary regularity assumptions, but the manuscript does not provide counterexamples, simulation evidence, or scope analysis showing these conditions hold beyond smooth unimodal cases (e.g., multimodal densities or non-Lipschitz boundaries); this is load-bearing for the bridge between paradigms asserted in the abstract and §5.

    Authors: We acknowledge that the asymptotic compatibility result in Theorem 5.1 is established under the stated local empirical process and boundary regularity assumptions, which are necessary for the quantitative convergence rates to Bayesian predictive density level sets. The manuscript does not include simulations or counterexamples for multimodal densities or non-Lipschitz boundaries, as the focus is on the theoretical derivation. These assumptions are standard for such asymptotic analyses and are explicitly listed. In revision we will add a dedicated paragraph in §5 discussing the scope, noting that the bridge to Bayesian methods holds when regularity is satisfied and identifying multimodal or irregular boundary cases as directions for future investigation. This clarifies the claims without overstating generality. revision: partial

  2. Referee: [§4.2] §4.2 (commuting diagram): while the decomposition into predictive distribution extraction and region derivation is presented as functorial, the proof does not explicitly verify that the diagram commutes while preserving the finite-sample coverage guarantee of CP; an explicit natural transformation or worked example with a standard nonconformity score (e.g., absolute residual) is needed to confirm the construction is not merely formal.

    Authors: We agree that an explicit verification strengthens the presentation. The commuting diagram is constructed so that the finite-sample coverage guarantee is preserved by the region derivation step, which inherits the conformal property independently of the predictive distribution extraction. To make this concrete, we will add a worked example in the revised §4.2 using the absolute residual nonconformity score. The example will exhibit the natural transformation explicitly, verify diagram commutation, and confirm that the coverage guarantee is maintained throughout, demonstrating that the decomposition is substantive. revision: yes

Circularity Check

0 steps flagged

No circularity: category-theoretic decomposition and asymptotic results are independently derived from definitions and stated assumptions.

full rationale

The paper constructs a category-theoretic representation of Full Conformal Prediction as a morphism in categories for stability and measurability, proves a commuting diagram that decomposes the procedure into extracting predictive distributions followed by region derivation, and establishes asymptotic convergence to Bayesian level sets under explicitly stated local empirical process and boundary regularity conditions. These steps rely on functorial properties and standard measure-theoretic arguments applied to the conformal construction, without any self-definitional loops, fitted parameters renamed as predictions, or load-bearing self-citations. The assumptions are invoked as premises for the compatibility result rather than being derived from the result itself, keeping the chain self-contained and externally verifiable against category theory and empirical process theory.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard category-theoretic axioms for morphisms, functors, and commuting diagrams together with domain assumptions on data regularity; no free parameters or invented entities are introduced.

axioms (2)
  • standard math Category theory supplies well-defined morphisms for stability of set-valued procedures and measurability of random regions.
    Invoked when representing Full Conformal Prediction as a morphism in two categories.
  • domain assumption Local empirical process and boundary regularity conditions hold for the data-generating process.
    Required for the quantitative rates in the asymptotic compatibility result.

pith-pipeline@v0.9.0 · 5758 in / 1417 out tokens · 36959 ms · 2026-05-19T06:02:55.757160+00:00 · methodology

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

71 extracted references · 71 canonical work pages

  1. [1]

    J., and Moral, S

    Abellán, J., Klir, G. J., and Moral, S. (2006). Disaggregated total uncertainty measure for credal sets. International Journal of General Systems , 1(35):29--44

    work page 2006

  2. [2]

    E., Kaddar, Y., Karwowski, J., Moss, S., Roy, D., Staton, S., and Yang, H

    Ackerman, N., Freer, C. E., Kaddar, Y., Karwowski, J., Moss, S., Roy, D., Staton, S., and Yang, H. (2024). Probabilistic programming interfaces for random graphs: Markov categories, graphons, and nominal sets. Proceedings of the ACM on Programming Languages , 8(POPL):1819–1849

    work page 2024

  3. [3]

    Aliprantis, C. D. and Border, K. C. (2006). Infinite Dimensional Analysis: a Hitchhiker's Guide . Berlin : Springer, 3rd edition

    work page 2006

  4. [4]

    N., Barber, R

    Angelopoulos, A. N., Barber, R. F., and Bates, S. (2024). Theoretical foundations of conformal prediction

    work page 2024

  5. [5]

    P., De Cooman, G., and Troffaes, M

    Augustin, T., Coolen, F. P., De Cooman, G., and Troffaes, M. C. (2014). Introduction to imprecise probabilities , volume 591. John Wiley & Sons

    work page 2014

  6. [6]

    F., Candes, E

    Barber, R. F., Candes, E. J., Ramdas, A., and Tibshirani, R. J. (2023). Conformal prediction beyond exchangeability. The Annals of Statistics , 51(2):816--845

    work page 2023

  7. [7]

    Bell, W. C. and Hagood, J. W. (1988). Separation properties and exact radon--nikod \'y m derivatives for bounded finitely additive measures. Pacific Journal of Mathematics , 131(2):237--248

    work page 1988

  8. [8]

    and Perrone, P

    Bohinen, M. and Perrone, P. (2025). Categorical algebra of conditional probability

    work page 2025

  9. [9]

    B\" o rgers, T. (1991). Upper hemicontinuity of the correspondence of subgame-perfect equilibrium outcomes. Journal of Mathematical Economics , 20(1):89--106

    work page 1991

  10. [10]

    Cabezas, L. M. C., Santos, V. S., Ramos, T. R., and Izbicki, R. (2025). Epistemic uncertainty in conformal scores: A unified approach

    work page 2025

  11. [11]

    Caprio, M. (2024). Imprecise Markov Semigroups and their Ergodicity . arXiv preprint arXiv:2405.00081 https://arxiv.org/abs/2405.00081

    work page arXiv 2024

  12. [12]

    Caprio, M. (2025). Optimal transport for -contaminated credal sets

    work page 2025

  13. [13]

    J., Lin, V., Ivanov, R., Sokolsky, O., and Lee, I

    Caprio, M., Dutta, S., Jang, K. J., Lin, V., Ivanov, R., Sokolsky, O., and Lee, I. (2024a). Credal Bayesian Deep Learning . Transactions on Machine Learning Research

    work page

  14. [14]

    and Gong, R

    Caprio, M. and Gong, R. (2023). Dynamic precise and imprecise probability kinematics. In Miranda, E., Montes, I., Quaeghebeur, E., and Vantaggi, B., editors, Proceedings of the Thirteenth International Symposium on Imprecise Probability: Theories and Applications , volume 215 of Proceedings of Machine Learning Research , pages 72--83. PMLR

    work page 2023

  15. [15]

    and Mukherjee, S

    Caprio, M. and Mukherjee, S. (2023). Ergodic theorems for dynamic imprecise probability kinematics. International Journal of Approximate Reasoning , 152:325--343

    work page 2023

  16. [16]

    Caprio, M., Sale, Y., H \"u llermeier, E., and Lee, I. (2024b). A Novel Bayes' Theorem for Upper Probabilities . In Cuzzolin, F. and Sultana, M., editors, Epistemic Uncertainty in Artificial Intelligence , pages 1--12, Cham. Springer Nature Switzerland

    work page

  17. [17]

    Caprio, M., Sale, Y., and Hüllermeier, E. (2025a). Conformal Prediction Regions are Imprecise Highest Density Regions . Accepted to ISIPTA 2025

    work page 2025

  18. [18]

    and Seidenfeld, T

    Caprio, M. and Seidenfeld, T. (2023). Constriction for sets of probabilities. In International Symposium on Imprecise Probability: Theories and Applications , pages 84--95. PMLR

    work page 2023

  19. [19]

    Caprio, M., Stutz, D., Li, S., and Doucet, A. (2025b). Conformalized credal regions for classification with ambiguous ground truth. Transactions on Machine Learning Research

    work page

  20. [20]

    Caprio, M., Sultana, M., Elia, E., and Cuzzolin, F. (2024c). Credal Learning Theory . NeurIPS 2024

    work page 2024

  21. [21]

    and Martin, R

    Cella, L. and Martin, R. (2021). Valid inferential models for prediction in supervised learning problems. In International Symposium on Imprecise Probability: Theories and Applications , pages 72--82. PMLR

    work page 2021

  22. [22]

    and Martin, R

    Cella, L. and Martin, R. (2022). Validity, consonant plausibility measures, and conformal prediction. International Journal of Approximate Reasoning , 141:110--130

    work page 2022

  23. [23]

    and Martin, R

    Cella, L. and Martin, R. (2023). Possibility-theoretic statistical inference offers performance and probativeness assurances. International Journal of Approximate Reasoning , 163:109060

    work page 2023

  24. [24]

    L., Caprio, M., and Muandet, K

    Chau, S. L., Caprio, M., and Muandet, K. (2025). Integral imprecise probability metrics

    work page 2025

  25. [25]

    Coolen, F. P. A. (1992). Imprecise highest density regions related to intervals of measures. Memorandum COSOR; Volume 9254

    work page 1992

  26. [26]

    Cornish, R. (2025). Stochastic neural network symmetrisation in markov categories

    work page 2025

  27. [27]

    Cuzzolin, F. (2020). The geometry of uncertainty: the geometry of imprecise probabilities . Springer Nature

    work page 2020

  28. [28]

    Day, A. (1975). Filter monads, continuous lattices and closure systems. Canadian Journal of Mathematics , 27(1):50--59

    work page 1975

  29. [29]

    de Finetti, B. (1974). Theory of Probability , volume 1. New York : Wiley

    work page 1974

  30. [30]

    de Finetti, B. (1975). Theory of Probability , volume 2. New York : Wiley

    work page 1975

  31. [31]

    and Spalsbury, A

    Diestel, J. and Spalsbury, A. (2014). The Joys of Haar measure . American Mathematical Society

    work page 2014

  32. [32]

    J., Ruchkin, I., Sokolsky, O., and Lee, I

    Dutta, S., Caprio, M., Lin, V., Cleaveland, M., Jang, K. J., Ruchkin, I., Sokolsky, O., and Lee, I. (2024). Distributionally Robust Statistical Verification with Imprecise Neural Networks . Accepted to HSCC 2025

    work page 2024

  33. [33]

    and Perrone, P

    Ensarguet, N. and Perrone, P. (2024). Categorical probability spaces, ergodic decompositions, and transitions to equilibrium

    work page 2024

  34. [34]

    Fang, X., Li, J., Mulchandani, V., and Kim, J.-E. (2025). Trustworthy ai: Safety, bias, and privacy -- a survey

    work page 2025

  35. [35]

    and Holmes, C

    Fong, E. and Holmes, C. C. (2021). Conformal bayesian computation. In Ranzato, M., Beygelzimer, A., Dauphin, Y., Liang, P., and Vaughan, J. W., editors, Advances in Neural Information Processing Systems , volume 34, pages 18268--18279. Curran Associates, Inc

    work page 2021

  36. [36]

    Fritz, T., Gonda, T., Lorenzin, A., Perrone, P., and Mohammed, A. S. (2025). Empirical measures and strong laws of large numbers in categorical probability

    work page 2025

  37. [37]

    Garner, R. (2020). The Vietoris monad and weak distributive laws. Applied Categorical Structures , 28(2):339--354

    work page 2020

  38. [38]

    J., and Cand `es, E

    Gibbs, I., Cherian, J. J., and Cand \`e s, E. J. (2023). Conformal prediction with conditional guarantees. arXiv preprint arXiv:2305.12616

    work page arXiv 2023

  39. [39]

    Gr \"u nwald, P. D. (2023). The e-posterior. Philosophical Transactions of the Royal Society A , 381(2247):20220146

    work page 2023

  40. [40]

    F., Sale, Y., and H \"u llermeier, E

    Hanselle, J., Javanmardi, A., Oberkofler, T. F., Sale, Y., and H \"u llermeier, E. (2025). Conformal prediction without nonconformity scores. In The 41st Conference on Uncertainty in Artificial Intelligence

    work page 2025

  41. [41]

    Hoff, P. D. (2009). A First Course in B ayesian Statistical Methods . New York : Springer

    work page 2009

  42. [42]

    Hofman, P., Sale, Y., and Hüllermeier, E. (2024). Quantifying aleatoric and epistemic uncertainty with proper scoring rules

    work page 2024

  43. [43]

    and Waegeman, W

    H \"u llermeier, E. and Waegeman, W. (2021). Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods. Machine Learning , 110(3):457--506

    work page 2021

  44. [44]

    Javanmardi, A., Sale, Y., Hofman, P., and H\"ullermeier, E. (2023). Conformal prediction with partially labeled data. In Papadopoulos, H., Nguyen, K. A., Boström, H., and Carlsson, L., editors, Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications , volume 204 of Proceedings of Machine Learning Research , pages 2...

    work page 2023

  45. [45]

    Javanmardi, A., Stutz, D., and Hüllermeier, E. (2024). Conformalized credal set predictors. In Proceedings of the 38th International Conference on Neural Information Processing Systems . NeurIPS

    work page 2024

  46. [46]

    H., Thies, S

    Javanmardi, A., Zargarbashi, S. H., Thies, S. M. A. R., Waegeman, W., Bojchevski, A., and Hüllermeier, E. (2025). Optimal conformal prediction under epistemic uncertainty

    work page 2025

  47. [47]

    Kleisli, H. (1962). Homotopy theory in abelian categories. Canadian Journal of Mathematics , 14:139--169

    work page 1962

  48. [48]

    Kleisli, H. (1965). Every standard construction is induced by a pair of adjoint functors. Proceedings of the American Mathematical Society , 16(3):544--546

    work page 1965

  49. [49]

    Korolev, Y. (2022). Two-layer neural networks with values in a banach space. SIAM Journal on Mathematical Analysis , 54(6):6358--6389

    work page 2022

  50. [50]

    Levi, I. (1980). The Enterprise of Knowledge . London, UK : MIT Press

    work page 1980

  51. [51]

    and Staton, S

    Liell-Cock, J. and Staton, S. (2024). Compositional imprecise probability

    work page 2024

  52. [52]

    Liu, B., Lv, N., Guo, Y., and Li, Y. (2024). Recent advances on federated learning: A systematic survey. Neurocomputing , 597:128019

    work page 2024

  53. [53]

    Lu, P., Caprio, M., Eaton, E., and Lee, I. (2024). IBCL: Zero-shot Model Generation for Task Trade-offs in Continual Learning . arXiv preprint arXiv:2305.14782

    work page arXiv 2024

  54. [54]

    Martin, R. (2022). Valid and efficient imprecise-probabilistic inference with partial priors, ii. general framework. arXiv preprint arXiv:2211.14567

    work page arXiv 2022

  55. [55]

    Martin, R. (2025). An efficient monte carlo method for valid prior-free possibilistic statistical inference

    work page 2025

  56. [56]

    Matache, C., Moss, S., and Staton, S. (2022). Concrete categories and higher-order recursion: With applications including probability, differentiability, and full abstraction. In Proceedings of the 37th Annual ACM/IEEE Symposium on Logic in Computer Science , LICS ’22, page 1–14. ACM

    work page 2022

  57. [57]

    Michael, E. (1951). Topologies on spaces of subsets. Transactions of the American Mathematical Society , 71(1):152--182

    work page 1951

  58. [58]

    Papadopoulos, H. (2008). Inductive conformal prediction: Theory and application to neural networks. In Tools in artificial intelligence . Citeseer

    work page 2008

  59. [59]

    Perrone, P. (2024). Starting Category Theory . World Scientific

    work page 2024

  60. [60]

    Sale, Y., Bengs, V., Caprio, M., and H\" u llermeier, E. (2024). Second-order uncertainty quantification: A distance-based approach. In Salakhutdinov, R., Kolter, Z., Heller, K., Weller, A., Oliver, N., Scarlett, J., and Berkenkamp, F., editors, Proceedings of the 41st International Conference on Machine Learning , volume 235 of Proceedings of Machine Lea...

    work page 2024

  61. [61]

    Sale, Y., Caprio, M., and H\" u llermeier, E. (2023). Is the volume of a credal set a good measure for epistemic uncertainty? In Evans, R. J. and Shpitser, I., editors, Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence , volume 216 of Proceedings of Machine Learning Research , pages 1795--1804. PMLR

    work page 2023

  62. [62]

    and Vovk, V

    Shafer, G. and Vovk, V. (2008). A tutorial on conformal prediction. Journal of Machine Learning Research , 9(3)

    work page 2008

  63. [63]

    Shiebler, D., Gavranović, B., and Wilson, P. (2021). Category theory in machine learning

    work page 2021

  64. [64]

    Song, C. (2024). To describe or construct statistical learning models using the category-theoretical language. BIMSA Technical Report

    work page 2024

  65. [65]

    Stehlík, M. (2012). Category theory in statistical learning? Proceedings of the 33rd Linz Seminar of Fuzzy Set Theory , page 56

    work page 2012

  66. [66]

    C., Kumar, D., Tulabandhula, T., and Trivedi, A

    Stutts, A. C., Kumar, D., Tulabandhula, T., and Trivedi, A. (2024). Invited: Conformal inference meets evidential learning: Distribution-free uncertainty quantification with epistemic and aleatoric separability. In Proceedings of the 61st ACM/IEEE Design Automation Conference , DAC '24, New York, NY, USA. Association for Computing Machinery

    work page 2024

  67. [67]

    Troffaes, M. C. M. and de Cooman, G. (2014). Lower Previsions . Wiley Series in Probability and Statistics. John Wiley & Sons, Chichester, United Kingdom

    work page 2014

  68. [68]

    Vovk, V. (2013). Transductive conformal predictors. In Papadopoulos, H., Andreou, A. S., Iliadis, L., and Maglogiannis, I., editors, Artificial Intelligence Applications and Innovations , pages 348--360, Berlin, Heidelberg. Springer Berlin Heidelberg

    work page 2013

  69. [69]

    Vovk, V., Gammerman, A., and Shafer, G. (2005). Algorithmic learning in a random world , volume 29. Springer

    work page 2005

  70. [70]

    Walley, P. (1991). Statistical Reasoning with Imprecise Probabilities , volume 42 of Monographs on Statistics and Applied Probability . Chapman and Hall, London

    work page 1991

  71. [71]

    Wyler, O. (1981). Algebraic theories of continuous lattices. In Gierz, G., Hofmann, K. H., Keimel, K., Lawson, J. D., Mislove, M. W., and Scott, D. S., editors, Continuous Lattices , volume 871 of Lecture Notes in Mathematics , pages 390--413. Springer, Berlin

    work page 1981