mlsanityData to 14 Oct 2026

Connect the Dots: Tighter Discrete Approximations of Privacy Loss Distributions

arXiv 2207.04380 ↗First posted 10 Jul 2022cs.DS cs.CR cs.LGcited by 52

Not on the map. Why?

Abstract (arXiv)

The privacy loss distribution (PLD) provides a tight characterization of the privacy loss of a mechanism in the context of differential privacy (DP). Recent work has shown that PLD-based accounting allows for tighter (varepsilon, δ)-DP guarantees for many popular mechanisms compared to other known methods. A key question in PLD-based accounting is how to approximate any (potentially continuous) PLD with a PLD over any specified discrete support. We present a novel approach to this problem. Our approach supports both pessimistic estimation, which overestimates the hockey-stick divergence (i.e., δ) for any value of varepsilon, and optimistic estimation, which underestimates the hockey-stick divergence. Moreover, we show that our pessimistic estimate is the best possible among all pessimistic estimates. Experimental evaluation shows that our approach can work with much larger discretization intervals while keeping a similar error bound compared to previous approaches and yet give a better approximation than existing methods.

Key facts. Cited by 52 papers in the map (data through 14 Oct 2026). Builds on 14 of its 31 references that are in the map. First posted on arXiv 10 Jul 2022.

Builds on

14 of 31 references.

References · 14

  1. 2016Deep Learning with Differential Privacycited by 2,728
  2. 2017Renyi Differential Privacycited by 650
  3. 2016Concentrated Differential Privacy: Simplifications, Extensions, and Lower Boundscited by 346
  4. 2013The Composition Theorem for Differential Privacycited by 297
  5. 2014RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Responsecited by 272
  6. 2021Numerical Composition of Differential Privacycited by 180
  7. 2017Collecting Telemetry Data Privatelycited by 149
  8. 2021Optimal Accounting of Differential Privacy via Characteristic Functioncited by 86
  9. 2020Individual Privacy Accounting via a Renyi Filtercited by 59
  10. 2020Tight Differential Privacy for Discrete-Valued Mechanisms and for the Subsampled Gaussian Mechanism Using FFTcited by 43
  11. 2016Privacy Odometers and Filters: Pay-as-you-Go Compositioncited by 33
  12. 2021Fast and Memory Efficient Differentially Private-SGD via JL Projectionscited by 29
2 more
  1. 2021Computing Differential Privacy Guarantees for Heterogeneous Compositions Using FFTcited by 20
  2. 2021Practical Privacy Filters and Odometers with R'enyi Differential Privacy and Applications to Differentially Private Deep Learningcited by 12

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 3; 2023: 7; 2024: 19; 2025: 9; 2026: 14.

52 papers.

Sort:

By area:

  1. 2024Private Fine-tuning of Large Language Models with Zeroth-order Optimizationcited by 41
  2. 2024Fine-Tuning Large Language Models with User-Level Differential Privacycited by 29
  3. 2024Mind the Privacy Unit! User-Level Differential Privacy for Language Model Fine-Tuningcited by 24
  4. 2022Learning to Generate Image Embeddings with User-level Differential Privacycited by 23
  5. 2024How Private are DP-SGD Implementations?cited by 22
  6. 2024Tighter Privacy Auditing of DP-SGD in the Hidden State Threat Modelcited by 18
  7. 2023Unlocking Accuracy and Fairness in Differentially Private Image Classificationcited by 16
  8. 2024Scalable DP-SGD: Shuffling vs. Poisson Subsamplingcited by 16
  9. 2025VaultGemma: A Differentially Private Gemma Modelcited by 16
  10. 2023Bounding data reconstruction attacks with the hypothesis testing interpretation of differential privacycited by 15
  11. 2025Unifying Re-Identification, Attribute Inference, and Data Reconstruction Risks in Differential Privacycited by 15
  12. 2024Avoiding Pitfalls for Privacy Accounting of Subsampled Mechanisms under Compositioncited by 13
38 more
  1. 2023A Randomized Approach for Tight Privacy Accountingcited by 11
  2. 2024Subsampling is not Magic: Why Large Batch Sizes Work for Differentially Private Stochastic Optimisationcited by 11
  3. 2024Beyond the Calibration Point: Mechanism Comparison in Differential Privacycited by 9
  4. 2024A Hassle-free Algorithm for Private Learning in Practice: Don't Use Tree Aggregation, Use BLTscited by 9
  5. 2024pfl-research: simulation framework for accelerating research in Private Federated Learningcited by 8
  6. 2024Attack-Aware Noise Calibration for Differential Privacycited by 8
  7. 2022Private Ad Modeling with DP-SGDcited by 6
  8. 2024Auditing Differential Privacy Guarantees Using Density Estimationcited by 6
  9. 2025Gaussian DP for Reporting Differential Privacy Guarantees in Machine Learningcited by 6
  10. 2022The Saddle-Point Accountant for Differential Privacycited by 5
  11. 2023Sparsity-Preserving Differentially Private Training of Large Embedding Modelscited by 5
  12. 2025Sequentially Auditing Differential Privacycited by 5
  13. 2025Empirical Privacy Variancecited by 3
  14. 2025GeoClip: Geometry-Aware Clipping for Differentially Private SGDcited by 3
  15. 2025ACTG-ARL: Differentially Private Conditional Text Generation with RL-Boosted Controlcited by 3
  16. 2026Sampling-Free Privacy Accounting for Matrix Mechanisms under Random Allocationcited by 3
  17. 2023The importance of feature preprocessing for differentially private linear optimizationcited by 2
  18. 2024Tight Group-Level DP Guarantees for DP-SGD with Sampling via Mixture of Gaussians Mechanismscited by 2
  19. 2024Privacy Profiles for Private Selectioncited by 2
  20. 2024Closed-Form Bounds for DP-SGD against Record-level Inferencecited by 2
  21. 2026Less Random, More Private: What is the Optimal Subsampling Scheme for DP-SGD?cited by 2
  22. 2024Laplace Transform Interpretation of Differential Privacycited by 1
  23. 2024PAPAYA Federated Analytics Stack: Engineering Privacy, Scalability and Practicalitycited by 1
  24. 2025Private Rate-Constrained Optimization with Applications to Fair Learningcited by 1
  25. 2026Trade-off Functions for DP-SGD with Subsampling based on Random Allocation: Tight Upper and Lower Boundscited by 1
  26. 2026DP-MacAdam: Differentially Private Mechanism with Adaptive Clipping and Adaptive Momentumcited by 1
  27. 2023DPAF: Image Synthesis via Differentially Private Aggregation in Forward Phasecited by 0
  28. 2023Differentially Private Optimization for Non-Decomposable Objective Functionscited by 0
  29. 2024Adversarial Sample-Based Approach for Tighter Privacy Auditing in Final Model-Only Scenarioscited by 0
  30. 2025Secret-Protected Evolution for Differentially Private Synthetic Text Generationcited by 0
  31. 2026Decomposing Private Image Generation via Coarse-to-Fine Wavelet Modelingcited by 0
  32. 2026Lap2: Revisiting Laplace DP-SGD for High Dimensions via Majorization Theorycited by 0
  33. 2026Amplified Patch-Level Differential Privacy for Free via Random Croppingcited by 0
  34. 2026Differentially Private Model Mergingcited by 0
  35. 2026Provable Robustness against Backdoor Attacks via the Primal-Dual Perspective on Differential Privacycited by 0
  36. 2026Detectability in Diversity: Improved Canary Crafting for Privacy Auditing in One Runcited by 0
  37. 2026StraightDP: Geometry-Aware Differential Privacy for Rectified-Flow Transformerscited by 0
  38. 2026The Privacy-Hallucination Tradeoff in Differentially Private Language Modelscited by 0

50 most cited of 52 All 52 (JSON)