mlsanityData to 7 Oct 2026

A General Framework for Auditing Differentially Private Machine Learning

arXiv 2210.08643 ↗First posted 16 Oct 2022cs.LG cs.CRcited by 25

Not on the map. Why?

Abstract (arXiv)

We present a framework to statistically audit the privacy guarantee conferred by a differentially private machine learner in practice. While previous works have taken steps toward evaluating privacy loss through poisoning attacks or membership inference, they have been tailored to specific models or have demonstrated low statistical power. Our work develops a general methodology to empirically evaluate the privacy of differentially private machine learning implementations, combining improved privacy search and verification methods with a toolkit of influence-based poisoning attacks. We demonstrate significantly improved auditing power over previous approaches on a variety of models including logistic regression, Naive Bayes, and random forest. Our method can be used to detect privacy violations due to implementation errors or misuse. When violations are not present, it can aid in understanding the amount of information that can be leaked from a given dataset, algorithm, and privacy specification.

Key facts. Cited by 25 papers in the map (data through 7 Oct 2026). Builds on 12 of its 31 references that are in the map. First posted on arXiv 16 Oct 2022.

Builds on

12 of 31 references.

References · 12

  1. 2016Deep Learning with Differential Privacycited by 2,682
  2. 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
  3. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
  4. 2021Opacus: User-Friendly Differential Privacy Library in PyTorchcited by 375
  5. 2013The Composition Theorem for Differential Privacycited by 290
  6. 2020Auditing Differentially Private Machine Learning: How Private is Private SGD?cited by 213
  7. 2021Adversary Instantiation: Lower Bounds for Differentially Private Machine Learningcited by 193
  8. 2019Data Poisoning against Differentially-Private Learners: Attacks and Defensescited by 75
  9. 2019Diffprivlib: The IBM Differential Privacy Librarycited by 51
  10. 2018Detecting Violations of Differential Privacycited by 37
  11. 2016Differentially Private Random Decision Forests using Smooth Sensitivitycited by 16
  12. 2018Machine learning and genomics: precision medicine vs. patient privacycited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 10; 2024: 9; 2025: 4; 2026: 2.

25 papers.

Sort:

By area:

  1. 2023How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacycited by 150
  2. 2023Privacy Auditing with One (1) Training Runcited by 119
  3. 2023One-shot Empirical Privacy Estimation for Federated Learningcited by 35
  4. 2023PrivLM-Bench: A Multi-level Privacy Evaluation Benchmark for Language Modelscited by 35
  5. 2024Auditing f-Differential Privacy in One Runcited by 28
  6. 2023Unleashing the Power of Randomization in Auditing Differentially Private MLcited by 19
  7. 2024PANORAMIA: Privacy Auditing of Machine Learning Models without Retrainingcited by 17
  8. 2024Tighter Privacy Auditing of DP-SGD in the Hidden State Threat Modelcited by 17
  9. 2023A Randomized Approach for Tight Privacy Accountingcited by 11
  10. 2023Scaling Up Differentially Private LASSO Regularized Logistic Regression via Faster Frank-Wolfe Iterationscited by 8
  11. 2023A Note On Interpreting Canary Exposurecited by 6
  12. 2024Under manipulations, are some AI models harder to audit?cited by 5
13 more
  1. 2024Revisiting Differentially Private Hyper-parameter Tuningcited by 5
  2. 2025The Hitchhiker's Guide to Efficient, End-to-End, and Tight DP Auditingcited by 4
  3. 2023Probing the Transition to Dataset-Level Privacy in ML Models Using an Output-Specific and Data-Resolved Privacy Profilecited by 3
  4. 2024High-Dimensional Distributed Sparse Classification with Scalable Communication-Efficient Global Updatescited by 3
  5. 2023Epsilon*: Privacy Metric for Machine Learning Modelscited by 2
  6. 2025Stop Walking in Circles! Bailing Out Early in Projected Gradient Descentcited by 2
  7. 2024VFLGAN: Vertical Federated Learning-based Generative Adversarial Network for Vertically Partitioned Data Publicationcited by 1
  8. 2024VFLGAN-TS: Vertical Federated Learning-based Generative Adversarial Networks for Publication of Vertically Partitioned Time-Series Datacited by 1
  9. 2025MCMC for Bayesian estimation of Differential Privacy from Membership Inference Attackscited by 1
  10. 2024Position: Challenges and Opportunities for Differential Privacy in the U.S. Federal Governmentcited by 0
  11. 2025Towards Reliable and Generalizable Differentially Private Machine Learning (Extended Version)cited by 0
  12. 2026Protecting and Preserving Protest Dynamics for Responsible Analysiscited by 0
  13. 2026Detectability in Diversity: Improved Canary Crafting for Privacy Auditing in One Runcited by 0