Private Prediction Sets
Open accessThe DOI matches its Crossref record; 78 of the 82 references it cites were found in a registry or an index; none of the 15 sources its claims cite is checked for support yet.
Abstract
In real-world settings involving consequential decision-making, the deployment of machine learning systems generally requires both reliable uncertainty quantification and protection of individuals’ privacy. We present a framework that treats these two desiderata jointly. Our framework is based on conformal prediction, a methodology that augments predictive models to return prediction sets that provide uncertainty quantification—they provably cover the true response with a user-specified probability, such as 90%. One might hope that when used with privately trained models, conformal prediction would yield privacy guarantees for the resulting prediction sets; unfortunately this is not the case. To remedy this key problem, we develop a method that takes any pretrained predictive model and outputs differentially private prediction sets. Our method follows the general approach of split conformal prediction; we use holdout data to calibrate the size of the prediction sets but preserve privacy by using a privatized quantile subroutine. This subroutine compensates for the noise introduced to preserve privacy in order to guarantee correct coverage. We evaluate the method on large-scale computer vision data sets.
Overview
CoChat Summary
TL;DR
The paper develops a differentially private version of split conformal prediction that turns any pretrained predictive model plus a calibration set into prediction sets with finite-sample coverage and ε-differential privacy. It shows the cost of privacy in calibration is mild, with coverage upper bound scaling as Õ((nε)^−1) versus O(n^−1) for nonprivate conformal prediction.
Key findings
- The proposed private prediction-set procedure is ε-differentially private and satisfies coverage \(1-\alpha \le P\{Y \in C(X)\} \le 1-\alpha + O((n\epsilon)^{-1})\) (Theorem 1 informal preview).
- A simplified upper bound gives \(P\{Y \in C(X)\} \le 1-\alpha + O(\log(n\epsilon/\alpha)/(n\epsilon))\), compared with \(1-\alpha + O(n^{-1})\) for nonprivate conformal prediction (Corollary 1).
- In CIFAR-10 experiments with \(n=5000\), \(\epsilon=8\), \(\delta=1e-5\), and \(\alpha=0.1\), the nonprivate model had 73% accuracy and the private model 67%; private conformal prediction caused only a minuscule increase in set size, while private model training caused a larger set-size increase.
- On ImageNet with a nonprivate ResNet-152, \(n=30000\) calibration points and 20000 evaluation images, coverage was nearly constant over three orders of magnitude of bin numbers and showed little sensitivity to \(\epsilon\); on CoronaHack, using a nonprivate ResNet-18 with 75% accuracy, median coverage was 90.4% at \(\alpha=0.1\).
Methods
- The method follows split conformal prediction: compute conformity scores \(s_i = S_f(X_i,Y_i)\) on held-out calibration data, choose a cutoff \(\hat{s}\), and output \(C(x)=\{y:S(x,y)\le \hat{s}\}\).
- Privacy is achieved by replacing the ordinary conformal quantile with a privatized quantile subroutine: scores are discretized into bins and the exponential mechanism is applied, extending the private median routine of Feldman and Steinke.
- Experiments evaluate image classification on CIFAR-10, ImageNet, and CoronaHack; the score used is \(S(x,y)=1-f(x)_y\), and private model training on CIFAR-10 uses DP-SGD via Opacus with \(\epsilon=8\), \(\delta=1e-5\).
Limitations
- The paper states that the coverage upper bound requires weak regularity on the score distribution, such as a continuous distribution with bounded density and no point masses; the lower bound holds for any distribution, and the paper notes the assumption could be enforced with tiebreaking noise.
- The paper states that private model training causes a substantial accuracy decrease and larger prediction sets, and concludes that the cost of privacy lies primarily in model fitting rather than private calibration.
- Evident from the text: the empirical evaluation is limited to image classification data sets (CIFAR-10, ImageNet, CoronaHack), and the private model training experiment is only on CIFAR-10; the method also requires a held-out calibration set not used to train \(f\), and the full pipeline is differentially private only if the underlying model is also trained differentially privately.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures







Verification
Claims the paper makes
0 Valid · 1 Not valid · 9 Not checked · 58 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The ResNet-18 had a final accuracy of 75% after fine-tuning.”
§ Private Prediction Sets > 4.4 COVID-19 diagnosis
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The nonprivate model's accuracy (73%) was significantly higher than that of the private model (67%).”
§ Private Prediction Sets > 4.1 Isolating the effects of private model training and private conformal prediction
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Then, the prediction sets in (2) with cutoff ŝ satisfy the coverage property in (1).”
§ Private Prediction Sets > 3 Algorithm and Guarantees
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Theorem 3 (Coverage upper bound).The prediction sets in (2) with ŝ is as in Theorem 2, satisfy the following coverage upper bound:”
§ Private Prediction Sets > 3 Algorithm and Guarantees
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“Proposition 1 (Privacy guarantee).Algorithm 3 is ϵ-differentially private.”
§ Private Prediction Sets > 3 Algorithm and Guarantees
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
33 more claims
Create a free account to see all 68 claims and to verify the 15 references still open.
- It may state two separate claims as one. 14
- Probably not a substantial claim. 16
- Borderline: one of the tests for a substantial claim is unclear. 1
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Private Prediction Sets > B.1 Proof of Theorem 2
“and note that by Lemma 1 and the fact that q ≥ 0.5, P{E} ≥ 1 -γα.”
§ Private Prediction Sets > 3 Algorithm and Guarantees
“As the number of samples grows and the privacy constraint relaxes, the procedure chooses a less conservative quantile, eventually approaching the limiting value 1 -α.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Private Prediction Sets > A Auxiliary results
“their ⌈nq⌉-th order statistic is stochastically lower bounded by the ⌈nq⌉-th order statistic of a uniform distribution, which follows the Beta(⌈nα⌉, n -⌈nα⌉ + 1) distribution.”
§ Private Prediction Sets > A Auxiliary results
“Z Beta follows the beta distribution Beta(⌈nq⌉, n -⌈nq⌉ + 1)”
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 1×
“Otherwise, we use a non-private model on the ImageNet dataset [62], to investigate the performance of our procedure in a more challenging setting with a large number of possible labels.”
MethodLearning Multiple Layers of Features from Tiny Images
Krizhevsky, A. (2024) · doi:10.57702/zp44cu3g · Cited by 23,395
Calibrating Noise to Sensitivity in Private Data Analysis
Dwork, C., McSherry, F., Nissim, K., & Smith, A. (2006) · Lecture Notes in Computer Science · doi:10.1007/11681878_14 · Cited by 7,272
Deep Learning with Differential Privacy
Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L. (2016) · Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/2976749.2978318 · Cited by 6,578 · cited 2×
“The private model is trained with private SGD [14], as implemented in the Opacus library, with privacy parameters ϵ = 8 and δ = 1 e − 5.”
BackgroundMethodThe Algorithmic Foundations of Differential Privacy
Dwork, C., & Roth, A. (2014) · Foundations and Trends® in Theoretical Computer Science · doi:10.1561/0400000042 · Cited by 5,557
The Algorithmic Foundations of Differential Privacy
Dwork, C., & Roth, A. (2013) · doi:10.1561/9781601988195 · Cited by 3,856
ImageNet: A large-scale hierarchical image database
Deng, J. H., Dong, W. F., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvprw.2009.5206848 · Cited by 2,714
RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response
Erlingsson, Ú., Pihur, V., & Korolova, A. (2014) · Proceedings of the 2014 ACM SIGSAC Conference on Computer and Communications Security · doi:10.1145/2660267.2660348 · Cited by 1,838 · cited 1×
“Differential privacy [5] has become the de facto standard for privacy-preserving data analysis, as witnessed by its widespread adoption in large-scale systems such as those by Google [6, 7], Apple [8], Microsoft [9], and the US Census Bureau [10, 11].”
BackgroundMechanism Design via Differential Privacy
McSherry, F., & Talwar, K. (2007) · 48th Annual IEEE Symposium on Foundations of Computer Science (FOCS'07) · doi:10.1109/focs.2007.66 · Cited by 1,371
Towards Sharp Inapproximability For Any 2-CSP
Austrin, P. (2007) · 48th Annual IEEE Symposium on Foundations of Computer Science (FOCS'07) · doi:10.1109/focs.2007.41 · Cited by 1,294
Algorithmic Learning in a Random World
(2005) · doi:10.1007/b106715 · Cited by 1,119
Distribution-Free Predictive Inference for Regression
Lei, J., G’Sell, M., Rinaldo, A., Tibshirani, R. J., & Wasserman, L. (2018) · Journal of the American Statistical Association · doi:10.1080/01621459.2017.1307116 · Cited by 1,049 · cited 3×
“In non-private conformal prediction, the upper bound is 1 − α + O ( n − 1 ) [4].”
BackgroundMethod(2011) · PubMed · Cited by 990 · cited 1×
“This increasing adoption of differential privacy goes hand in hand with steady progress in differentially private model training, ranging across both convex [12, 13] and non-convex [14, 15] settings.”
BackgroundDetermination of Sample Sizes for Setting Tolerance Limits
Wilks, S. S. (1941) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177731788 · Cited by 851 · cited 1×
“Prediction sets as a way to represent uncertainty are a classical idea, going back at least to tolerance regions in the 1940s [24, 25, 26, 27].”
BackgroundA tutorial on conformal prediction
Shafer, G., & Vovk, V. (2007) · arXiv · doi:10.48550/arxiv.0706.3188 · Cited by 655 · cited 2×
“Conformal prediction [30, 3, 31] is a related way of producing predictive sets with finite-sample guarantees.”
BackgroundPrivate Empirical Risk Minimization: Efficient Algorithms and Tight Error Bounds
Bassily, R., Smith, A., & Thakurta, A. (2014) · 2014 IEEE 55th Annual Symposium on Foundations of Computer Science · doi:10.1109/focs.2014.56 · Cited by 621 · cited 1×
“This increasing adoption of differential privacy goes hand in hand with steady progress in differentially private model training, ranging across both convex [12, 13] and non-convex [14, 15] settings.”
BackgroundProchlo: Strong Privacy for Analytics in the Crowd
Bittau, A., Erlingsson, Ú., Maniatis, P., Mironov, I., Raghunathan, A., Lie, D., Rudominer, M., Kode, U., Tinnes, J., & Seefeld, B. (2017) · Proceedings of the 26th Symposium on Operating Systems Principles · doi:10.1145/3132747.3132769 · Cited by 378
Predictive inference with the jackknife+
Barber, R. F., Candès, E. J., Ramdas, A., & Tibshirani, R. J. (2021) · The Annals of Statistics · doi:10.1214/20-aos1965 · Cited by 351 · cited 1×
“[34] refine this approach to re-use data for both training and calibration, improving statistical efficiency.”
MethodStatistical Tolerance Regions: Theory, Applications, and Computation
Krishnamoorthy, K., & Mathew, T. (2009) · Wiley Series in Probability and Statistics · doi:10.1002/9780470473900 · Cited by 349 · cited 2×
“See Krishnamoorthy & Mathew [28] for an overview of tolerance regions and Park et al. [29] for a recent application to deep learning models.”
BackgroundInductive Confidence Machines for Regression
Papadopoulos, H., Proedrou, K., Vovk, V., & Gammerman, A. (2002) · Lecture Notes in Computer Science · doi:10.1007/3-540-36755-1_29 · Cited by 346 · cited 2×
“In order to generate prediction sets satisfying this property, we will use ideas from split conformal prediction [2, 3, 4], modifying this approach to ensure privacy.”
BackgroundMethodStatistical Prediction with Special Reference to the Problem of Tolerance Limits
Wilks, S. S. (1942) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177731537 · Cited by 334 · cited 1×
“Prediction sets as a way to represent uncertainty are a classical idea, going back at least to tolerance regions in the 1940s [24, 25, 26, 27].”
BackgroundThe U.S. Census Bureau Adopts Differential Privacy
Abowd, J. M. (2018) · Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining · doi:10.1145/3219819.3226070 · Cited by 292 · cited 1×
“Differential privacy [5] has become the de facto standard for privacy-preserving data analysis, as witnessed by its widespread adoption in large-scale systems such as those by Google [6, 7], Apple [8], Microsoft [9], and the US Census Bureau [10, 11].”
BackgroundCollecting Telemetry Data Privately
Ding, B., Kulkarni, J., & Yekhanin, S. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1712.01524 · Cited by 254 · cited 1×
“Differential privacy [5] has become the de facto standard for privacy-preserving data analysis, as witnessed by its widespread adoption in large-scale systems such as those by Google [6, 7], Apple [8], Microsoft [9], and the US Census Bureau [10, 11].”
BackgroundPrivacy-preserving statistical estimation with optimal convergence rates
Smith, A. (2011) · Proceedings of the forty-third annual ACM symposium on Theory of computing · doi:10.1145/1993636.1993743 · Cited by 251 · cited 1×
“At a technical level, closest to our algorithm on the privacy side are existing methods for reporting histograms and quantiles in a privacy-preserving fashion [5, 16, 17, 18, 19].”
MethodA gentle introduction to conformal prediction and distribution-free uncertainty quantification
Angelopoulos, A. N., & Bates, S. (2021) · arXiv · doi:10.48550/arxiv.2107.07511 · Cited by 240 · cited 1×
“We suggest [54] and [31] as introductory tutorials on conformal prediction for the unfamiliar reader.”
BackgroundSelf-citationConformalized Quantile Regression
Romano, Y., Patterson, E., & Candès, E. J. (2019) · arXiv · doi:10.48550/arxiv.1905.03222 · Cited by 236 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundDifferentially private histogram publication
Xu, J., Zhang, Z., Xiao, X., Yang, Y., Yu, G., & Winslett, M. (2013) · The VLDB Journal · doi:10.1007/s00778-013-0309-y · Cited by 228 · cited 1×
“At a technical level, closest to our algorithm on the privacy side are existing methods for reporting histograms and quantiles in a privacy-preserving fashion [5, 16, 17, 18, 19].”
MethodConditional validity of inductive conformal predictors
Vovk, V. (2013) · Machine Learning · doi:10.1007/s10994-013-5355-6 · Cited by 225 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundMechanism Design via Differential Privacy
McSherry, F., & Talwar, K. (2007) · Annual Symposium on Foundations of Computer Science · doi:10.1109/focs.2007.4389483 · Cited by 202
An Extension of Wilks' Method for Setting Tolerance Limits
Wald, A. (1943) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177731491 · Cited by 193 · cited 1×
“Prediction sets as a way to represent uncertainty are a classical idea, going back at least to tolerance regions in the 1940s [24, 25, 26, 27].”
BackgroundHappyMap : A Generalized Multicalibration Method
Poujade, K., Travé-Massuyès, L., Pirard, J., & Vieillevigne, L. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itcs.2023.41 · Cited by 189
Finite Sample Differentially Private Confidence Intervals
Karwa, V., & Vadhan, S. (2017) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itcs.2018.44 · Cited by 187
Tukey, J. W. (1947) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177730343 · Cited by 173 · cited 1×
“Prediction sets as a way to represent uncertainty are a classical idea, going back at least to tolerance regions in the 1940s [24, 25, 26, 27].”
BackgroundVovk, V. (2015) · Annals of Mathematics and Artificial Intelligence · doi:10.1007/s10472-013-9368-4 · Cited by 166 · cited 1×
“Vovk [33] and Barber et al. [34] refine this approach to re-use data for both training and calibration, improving statistical efficiency.”
MethodConformal Inference of Counterfactuals and Individual Treatment Effects
Lei, L., & Candès, E. J. (2021) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/rssb.12445 · Cited by 166
Machine-Learning Applications of Algorithmic Randomness
Vovk, V., Gammerman, A., & Saunders, C. J. (1999) · ePrints Soton (University of Southampton) · Cited by 163
Robust Validation: Confident Predictions Even When Distributions Shift
Cauchois, M., Gupta, S., Ali, A., & Duchi, J. C. (2024) · Journal of the American Statistical Association · doi:10.1080/01621459.2023.2298037 · Cited by 160 · cited 1×
“Further extensions address problems in distribution estimation [47, 48], handling or testing distribution shift [49, 50, 51], causal inference [52], and controlling other notions of statistical error [53].”
BackgroundLeast Ambiguous Set-Valued Classifiers With Bounded Error Levels
Sadinle, M., Lei, J., & Wasserman, L. (2019) · Journal of the American Statistical Association · doi:10.1080/01621459.2017.1395341 · Cited by 159 · cited 2×
“We seek to create private prediction sets, C ( X i ) ⊆ { 1 , ..., K } , achieving coverage as in equation (1), using the following score function: as in Sadinle et al. [35].”
BackgroundMethodThe limits of distribution-free conditional predictive inference
Foygel Barber, R., Candès, E. J., Ramdas, A., & Tibshirani, R. J. (2021) · Information and Inference: A Journal of the IMA · doi:10.1093/imaiai/iaaa017 · Cited by 147 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundDistribution-free, Risk-controlling Prediction Sets
Bates, S., Angelopoulos, A., Lei, L., Malik, J., & Jordan, M. (2021) · Journal of the ACM · doi:10.1145/3478535 · Cited by 144 · cited 1×
“Further extensions address problems in distribution estimation [47, 48], handling or testing distribution shift [49, 50, 51], causal inference [52], and controlling other notions of statistical error [53].”
BackgroundSelf-citationRAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response
Erlingsson, Ú., Pihur, V., & Korolova, A. (2014) · doi:10.48550/arxiv.1407.6981 · Cited by 131 · cited 1×
“Differential privacy [5] has become the de facto standard for privacy-preserving data analysis, as witnessed by its widespread adoption in large-scale systems such as those by Google [6, 7], Apple [8], Microsoft [9], and the US Census Bureau [10, 11].”
BackgroundDifferentially Private Histogram Publication
Xu, J., Zhang, Z., Xiao, X., Yang, Y., & Yu, G. (2012) · 2012 IEEE 28th International Conference on Data Engineering · doi:10.1109/icde.2012.48 · Cited by 129
Differentially Private M-Estimators
Lei, L., & Jing, J. (2011) · Cited by 96
Nonparametric predictive distributions based on conformal prediction
Vovk, V., Shen, J., Manokhin, V., & Xie, M.-G. (2018) · Machine Learning · doi:10.1007/s10994-018-5755-8 · Cited by 91 · cited 1×
“Further extensions address problems in distribution estimation [47, 48], handling or testing distribution shift [49, 50, 51], causal inference [52], and controlling other notions of statistical error [53].”
BackgroundMoment Multicalibration for Uncertainty EstimationPreprint
Jung, C., Lee, C., Pai, M. M., Roth, A., & Vohra, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.08037 · Cited by 89 · cited 1×
“First, the calibration technique in Jung et al. [55] and Gupta et al. [56] generates prediction sets via the estimation of higher moments across many overlapping sub-populations.”
MethodPAC Confidence Sets for Deep Neural Networks via Calibrated PredictionPreprint
Park, S., Bastani, O., Matni, N., & Lee, I. (2019) · International Conference on Learning Representations · doi:10.48550/arxiv.2001.00106 · Cited by 84 · cited 1×
“See Krishnamoorthy & Mathew [28] for an overview of tolerance regions and Park et al. [29] for a recent application to deep learning models.”
BackgroundA conformal prediction approach to explore functional data
Lei, J., Rinaldo, A., & Wasserman, L. (2015) · Annals of Mathematics and Artificial Intelligence · doi:10.1007/s10472-013-9366-6 · Cited by 83 · cited 1×
“Most relevant to the present work, split conformal prediction [2, 32, 4] is a convenient version that uses data splitting to give prediction sets in a computationally efficient way.”
BackgroundUncertainty Sets for Image Classifiers using Conformal Prediction
Angelopoulos, A., Bates, S., Malik, J., & Jordan, M. I. (2020) · arXiv · doi:10.48550/arxiv.2009.14193 · Cited by 79 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundSelf-citationFlexible distribution-free conditional predictive bands using density estimatorsPreprint
Izbicki, R., Shimizu, G. T., & Stern, R. B. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.05575 · Cited by 77 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundClassification with confidence
Lei, J. (2014) · Biometrika · doi:10.1093/biomet/asu038 · Cited by 68 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundEfficient set-valued prediction in multi-class classification
Mortier, T., Wydmuch, M., Dembczyński, K., Hüllermeier, E., & Waegeman, W. (2021) · Data Mining and Knowledge Discovery · doi:10.1007/s10618-021-00751-x · Cited by 68 · cited 1×
“Second, there is a family of techniques that define a utility function balancing set-size and coverage and then search for set-valued predictors to maximize this utility [57, 58, 59].”
BackgroundDifferentially Private Ordinary Least Squares
Sheffet, O. (2019) · Journal of Privacy and Confidentiality · doi:10.29012/jpc.654 · Cited by 63 · cited 1×
“Finally, there have also been significant efforts to quantify uncertainty with formal privacy guarantees through various types of private confidence intervals [20, 21, 22, 23].”
BackgroundLearning Nondeterministic Classifiers
del Coz, J. J., Díez, J., & Bahamonde, A. (2009) · Journal of Machine Learning Research · doi:10.5555/1577069.1755862 · Cited by 60 · cited 1×
“Second, there is a family of techniques that define a utility function balancing set-size and coverage and then search for set-valued predictors to maximize this utility [57, 58, 59].”
BackgroundLocally Private Mean Estimation: Z-test and Tight Confidence IntervalsPreprint
Gaboardi, M., Rogers, R., & Sheffet, O. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.08054 · Cited by 48 · cited 1×
“Finally, there have also been significant efforts to quantify uncertainty with formal privacy guarantees through various types of private confidence intervals [20, 21, 22, 23].”
BackgroundGeneralization for Adaptively-chosen Estimators via Stable MedianPreprint
Feldman, V., & Steinke, T. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1706.05069 · Cited by 43 · cited 3×
“The following is an extension of the the analogous result for the private median due to Feldman and Steinke [19].”
MethodResultPrediction and Outlier Detection in Classification Problems
Guan, L., & Tibshirani, R. (2022) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/rssb.12443 · Cited by 43 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundConformal prediction with localizationPreprint
Guan, L. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.08558 · Cited by 40 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundDistribution-Free Predictive Inference For Regression
Lei, J., G’Sell, M., Rinaldo, A., Tibshirani, R. J., & Wasserman, L. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1604.04173 · Cited by 38 · cited 3×
“In non-private conformal prediction, the upper bound is 1 − α + O ( n − 1 ) [4].”
BackgroundMethodConformal Inference of Counterfactuals and Individual Treatment EffectsPreprint
Lei, L., & Candès, E. J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.06138 · Cited by 32
Differentially Private Confidence Intervals for Empirical Risk Minimization
Wang, Y., Kifer, D., & Lee, J. (2019) · Journal of Privacy and Confidentiality · doi:10.29012/jpc.660 · Cited by 31
Knowing what you know: valid and validated confidence sets in multiclass and multilabel prediction
Cauchois, M., Gupta, S., & Duchi, J. (2020) · arXiv · doi:10.48550/arxiv.2004.10181 · Cited by 31
Online Multivalid Learning: Means, Moments, and Prediction Intervals
Gupta, V., Jung, C., Noarov, G., Pai, M. M., & Roth, A. (2021) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itcs.2022.82 · Cited by 30
Differential Privacy and the US Census
Dwork, C. (2019) · ACM SIGMOD-SIGACT-SIGAI Symposium on Principles of Database Systems (PODS) · doi:10.1145/3294052.3322188 · Cited by 29 · cited 1×
“Differential privacy [5] has become the de facto standard for privacy-preserving data analysis, as witnessed by its widespread adoption in large-scale systems such as those by Google [6, 7], Apple [8], Microsoft [9], and the US Census Bureau [10, 11].”
BackgroundHechtlinger, Y., Póczos, B., & Wasserman, L. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.09460 · Cited by 28 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundClassification with Valid and Adaptive Coverage
Romano, Y., Sesia, M., & Candès, E. J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.02544 · Cited by 23 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundA Distribution-Free Test of Covariate Shift Using Conformal PredictionPreprint
Hu, X., & Lei, J. (2020) · arXiv (Cornell University) · Cited by 21
Make Up Your Mind: The Price of Online Queries in Differential Privacy
Bun, M., Steinke, T., & Ullman, J. (2017) · ACM-SIAM Symposium on Discrete Algorithms (SODA) · doi:10.1137/1.9781611974782.85 · Cited by 17
Oracle Efficient Private Non-Convex OptimizationPreprint
Neel, S., Roth, A., Vietri, G., & Wu, S. Y.-P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.01783 · Cited by 16
Conformal calibratorsPreprint
Vovk, V., Petej, I., Toccaceli, P., & Gammerman, A. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.06579 · Cited by 8 · cited 1×
“Further extensions address problems in distribution estimation [47, 48], handling or testing distribution shift [49, 50, 51], causal inference [52], and controlling other notions of statistical error [53].”
BackgroundNonparametric predictive distributions based on conformal prediction.
Vovk, V., Shen, J., Manokhin, V., & Xie, M. (2017) · Cited by 8
Classification with Set-Valued Decision Functions
Grycko, E. (1993) · Studies in classification, data analysis, and knowledge organization · doi:10.1007/978-3-642-50974-2_22 · Cited by 5 · cited 1×
“Second, there is a family of techniques that define a utility function balancing set-size and coverage and then search for set-valued predictors to maximize this utility [57, 58, 59].”
BackgroundPrediction and outlier detection in classification problemsPreprint
Guan, L., & Tibshirani, R. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.04396 · Cited by 5 · cited 1×
“Recent work has targeted desiderata such as small set sizes [35, 36], coverage that is approximately balanced across feature space [37, 38, 39, 40, 41, 42, 43], and coverage that is balanced across classes [44, 35, 45, 46].”
BackgroundMake Up Your Mind: The Price of Online Queries in Differential PrivacyPreprint
Bun, M., Steinke, T., & Ullman, J. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1604.04618 · Cited by 5
Least Ambiguous Set-Valued Classifiers With Bounded Error Levels
Sadinle, M., Lei, J., & Wasserman, L. (2017) · Figshare · doi:10.6084/m9.figshare.5552596 · Cited by 5
Differentially Private Confidence Intervals for Empirical Risk MinimizationPreprint
Wang, Y., Kifer, D., & Lee, J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.03794 · Cited by 1 · cited 1×
“Finally, there have also been significant efforts to quantify uncertainty with formal privacy guarantees through various types of private confidence intervals [20, 21, 22, 23].”
BackgroundDatabiology Lab CORONAHACK: Collection of Public COVID-19 DataPreprint
Caballero-Pérez, J., Contreras, J. M. C., Pérez, C. D. B., & Alvarez, F. L. (2020) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2020.10.22.328864 · Cited by 1 · cited 2×
“This will allow an individual to contribute a [1] with their corresponding private prediction sets at α = 10% from a ResNet-18.”
BackgroundMethodConformal Prediction Under Covariate Shift
Tibshirani, R. J., Barber, R. F., Candes, E. J., & Ramdas, A. (2019) · arXiv · doi:10.48550/arxiv.1904.06019 · cited 1×
“Further extensions address problems in distribution estimation [47, 48], handling or testing distribution shift [49, 50, 51], causal inference [52], and controlling other notions of statistical error [53].”
Background
Research works printed but not linked: 1
- [2]2. Abowd, J. M. (2018). The US census bureau adopts differential privacy. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining (pp. 2867-2867).
Cited by
Learn then Test: Calibrating Predictive Algorithms to Achieve Risk Control
Angelopoulos, A. N., Bates, S., Candès, E. J., Jordan, M. I., & Lei, L. (2021) · arXiv · doi:10.48550/arxiv.2110.01052 · Cited by 100
“Recent extensions also address topics such as distribution estimation [27], causal inference [28], sur-2 vival analysis [29], differential privacy [30], outlier detection [31], speeding up the test-time evaluation of complex models [32, 33], the few-shot setting [34], handling dependent data [35,…”
Self-citationLearn then test: Calibrating predictive algorithms to achieve risk control
Angelopoulos, A. N., Bates, S., Candès, E. J., Jordan, M. I., & Lei, L. (2025) · The Annals of Applied Statistics · doi:10.1214/24-aoas1998 · Cited by 466
Self-citationLu, C., Angelopoulos, A. N., & Pomerantz, S. R. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-16452-1_52 · Cited by 71
Self-citationZhu, M., Zecchin, M., Park, S., Guo, C., Feng, C., & Simeone, O. (2024) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2024.3358615 · Cited by 36
“Setting Following the experimental setting in reference [37], we use the CoronaHack data set, a public chest X-ray data set involving 5908 images classified using C = 3 labels, namely normal, viral pneumonia, and bacterial pneumonia.”
MethodGhoshal, B., Woof, W. A., Mendes, B. S., Al‐Khuzaei, S., de Guimarães, T. A. C., Varela, M. D., Liu, Y., Sen, S., Lin, S., Shah, M., et al. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73158-7_5 · Cited by 2
Metadata
- Authors
- Angelopoulos, A. N., Bates, S., Zrnic, T., & Jordan, M. I.
- Affiliations
- University of California, Berkeley (US)
- Venue
- Harvard Data Science Review
- Publisher
- MIT Press
- Published
- 2022-04-28
- Volume
- 4
- Issue
- 2
- ISSN
- 2644-2353 (ISSN-L)
- 2688-8513
- OpenAlex
- W3129027643
- Cited by
- 8
- References
- 82
- Citations by year
- 2026: 3 · 2025: 1 · 2024: 2 · 2023: 1 · 2022: 1
- FWCI
- 0.41
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by
- Locations
- Harvard Data Science Review (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- conformal prediction, split conformal prediction, differential privacy, prediction sets, uncertainty quantification, privacy-preserving machine learning, coverage guarantees, computer vision
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.