mlsanityData to 7 Oct 2026

Privacy Amplification by Subsampling: Tight Analyses via Couplings and Divergences

arXiv 1807.01647 ↗First posted 4 Jul 2018cs.LG cs.CR stat.MLcited by 237

Not on the map. Why?

Abstract (arXiv)

Differential privacy comes equipped with multiple analytical tools for the design of private data analyses. One important tool is the so-called "privacy amplification by subsampling" principle, which ensures that a differentially private mechanism run on a random subsample of a population provides higher privacy guarantees than when run on the entire population. Several instances of this principle have been studied for different random subsampling methods, each with an ad-hoc analysis. In this paper we present a general method that recovers and improves prior analyses, yields lower bounds and derives new instances of privacy amplification by subsampling. Our method leverages a characterization of differential privacy as a divergence which emerged in the program verification community. Furthermore, it introduces new tools, including advanced joint convexity and privacy profiles, which might be of independent interest.

Key facts. Cited by 237 papers in the map (data through 7 Oct 2026). First posted on arXiv 4 Jul 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 30; 2022: 40; 2023: 54; 2024: 38; 2025: 38; 2026: 37.

237 papers.

Sort:

By area:

  1. 2021Adversary Instantiation: Lower Bounds for Differentially Private Machine Learningcited by 193
  2. 2023How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacycited by 150
  3. 2021Machine Unlearning via Algorithmic Stabilitycited by 128
  4. 2021Large-Scale Differentially Private BERTcited by 117
  5. 2021Hyperparameter Tuning with Renyi Differential Privacycited by 104
  6. 2023Privacy-Preserving In-Context Learning with Differentially Private Few-Shot Generationcited by 94
  7. 2021Do Not Let Privacy Overbill Utility: Gradient Embedding Perturbation for Private Learningcited by 86
  8. 2021Optimal Accounting of Differential Privacy via Characteristic Functioncited by 83
  9. 2022DisPFL: Towards Communication-Efficient Personalized Federated Learning via Decentralized Sparse Trainingcited by 74
  10. 2021Differential Privacy Dynamics of Langevin Diffusion and Noisy Gradient Descentcited by 71
  11. 2021Selective Differential Privacy for Language Modelingcited by 61
  12. 2021Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergencecited by 54
38 more
  1. 2023Privacy-Preserving In-Context Learning for Large Language Modelscited by 50
  2. 2022When Does Differentially Private Learning Not Suffer in High Dimensions?cited by 48
  3. 2022Composition of Differential Privacy & Privacy Amplification by Subsamplingcited by 42
  4. 2024Private Fine-tuning of Large Language Models with Zeroth-order Optimizationcited by 41
  5. 2021Releasing Graph Neural Networks with Differential Privacy Guaranteescited by 32
  6. 2022Just Fine-tune Twice: Selective Differential Privacy for Large Language Modelscited by 32
  7. 2022Sparse Random Networks for Communication-Efficient Federated Learningcited by 32
  8. 2023Conformal Prediction for Federated Uncertainty Quantification Under Label Shiftcited by 29
  9. 2024Fine-Tuning Large Language Models with User-Level Differential Privacycited by 28
  10. 2022Muffliato: Peer-to-Peer Privacy Amplification for Decentralized Optimization and Averagingcited by 27
  11. 2021The Role of Adaptive Optimizers for Honest Private Hyperparameter Selectioncited by 25
  12. 2024Decentralized Federated Learning: A Survey on Security and Privacycited by 25
  13. 2021Privacy Amplification for Federated Learning via User Sampling and Wireless Aggregationcited by 22
  14. 2023Privacy Amplification via Compression: Achieving the Optimal Privacy-Accuracy-Communication Trade-off in Distributed Mean Estimationcited by 22
  15. 2021Computing Differential Privacy Guarantees for Heterogeneous Compositions Using FFTcited by 20
  16. 2022Individual Privacy Accounting for Differentially Private Stochastic Gradient Descentcited by 20
  17. 2023Differentially Private Natural Language Models: Recent Advances and Future Directionscited by 20
  18. 2023Leave-one-out Distinguishability in Machine Learningcited by 20
  19. 2021Differentially Private Sliced Wasserstein Distancecited by 19
  20. 2023Practical Differentially Private Hyperparameter Tuning with Subsamplingcited by 19
  21. 2023Threshold KNN-Shapley: A Linear-Time and Privacy-Friendly Approach to Data Valuationcited by 19
  22. 2021Local Differential Privacy Is Equivalent to Contraction of Eγ-Divergencecited by 17
  23. 2021Wide Network Learning with Differential Privacycited by 17
  24. 2021Gradient Leakage Attack Resilient Deep Learningcited by 17
  25. 2023ReSQueing Parallel and Private Stochastic Convex Optimizationcited by 17
  26. 2022On the Privacy Properties of GAN-generated Samplescited by 16
  27. 2022~Optimal Differentially Private Learning of Thresholds and Quasi-Concave Optimizationcited by 16
  28. 2021Federated Intrusion Detection for IoT with Heterogeneous Cohort Privacycited by 14
  29. 2021On Convergence of Federated Averaging Langevin Dynamicscited by 14
  30. 2022Edgeworth Accountant: An Analytical Approach to Differential Privacy Compositioncited by 14
  31. 2023Bounding data reconstruction attacks with the hypothesis testing interpretation of differential privacycited by 14
  32. 2024Differentially Private Zeroth-Order Methods for Scalable Large Language Model Finetuningcited by 14
  33. 2022Provably Confidential Language Modellingcited by 13
  34. 2022DP2-VAE: Differentially Private Pre-trained Variational Autoencoderscited by 13
  35. 2022DPIS: An Enhanced Mechanism for Differentially Private SGD with Importance Samplingcited by 13
  36. 2023Securing Distributed SGD against Gradient Leakage Threatscited by 13
  37. 2023Graphical vs. Deep Generative Models: Measuring the Impact of Differentially Private Mechanisms and Budgets on Utilitycited by 13
  38. 2023Unified Enhancement of Privacy Bounds for Mixture Mechanisms via f-Differential Privacycited by 13

50 most cited of 237 All 237 (JSON)