mlsanityData to 7 Oct 2026

Differentially Private Query Release Through Adaptive Projection

arXiv 2103.06641 ↗First posted 11 Mar 2021cs.LG cs.CRcited by 50

Not on the map. Why?

Abstract (arXiv)

We propose, implement, and evaluate a new algorithm for releasing answers to very large numbers of statistical queries like k-way marginals, subject to differential privacy. Our algorithm makes adaptive use of a continuous relaxation of the Projection Mechanism, which answers queries on the private dataset using simple perturbation, and then attempts to find the synthetic dataset that most closely matches the noisy answers. We use a continuous relaxation of the synthetic dataset domain which makes the projection loss differentiable, and allows us to use efficient ML optimization techniques and tooling. Rather than answering all queries up front, we make judicious use of our privacy budget by iteratively and adaptively finding queries for which our (relaxed) synthetic data has high error, and then repeating the projection. We perform extensive experimental evaluations across a range of parameters and datasets, and find that our method outperforms existing algorithms in many cases, especially when the privacy budget is small or the query class is large.

Key facts. Cited by 50 papers in the map (data through 7 Oct 2026). Builds on 15 of its 44 references that are in the map. First posted on arXiv 11 Mar 2021.

Builds on

15 of 44 references.

References · 15

  1. 2016Concentrated Differential Privacy: Simplifications, Extensions, and Lower Boundscited by 339
  2. 2016From Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classificationcited by 337
  3. 2020DP-CGAN: Differentially Private Synthetic Data and Label Generationcited by 118
  4. 2019Graphical-model based estimation and inference for differential privacycited by 73
  5. 2020New Oracle-Efficient Algorithms for Private Synthetic Data Releasecited by 43
  6. 2019Practical Differentially Private Top-k Selection with Pay-what-you-get Compositioncited by 38
  7. 2018Optimizing error of high-dimensional statistical queries under differential privacycited by 29
  8. 2014Dual Query: Practical Private Query Release for High Dimensional Datacited by 26
  9. 2020P3GM: Private High-Dimensional Data Release via Privacy Preserving Phased Generative Modelcited by 21
  10. 2020Private Post-GAN Boostingcited by 20
  11. 2013Efficient Algorithms for Privately Releasing Marginals via Convex Relaxationscited by 17
  12. 2017Marginal Release Under Local Differential Privacycited by 9
3 more
  1. 2018Towards Instance-Optimal Private Query Releasecited by 9
  2. 2018How to Use Heuristics for Differential Privacycited by 6
  3. 2013Faster Private Release of Marginals on Small Databasescited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 3; 2022: 8; 2023: 13; 2024: 15; 2025: 8; 2026: 3.

50 papers.

Sort:

By area:

  1. 2023How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacycited by 150
  2. 2021Iterative Methods for Private Synthetic Data: Unifying Framework and New Methodscited by 58
  3. 2023SoK: Privacy-Preserving Data Synthesiscited by 37
  4. 2022Private Synthetic Data for Multitask Learning and Marginal Queriescited by 35
  5. 2024Privacy-Preserving Instructions for Aligning Large Language Modelscited by 34
  6. 2023A Linear Reconstruction Approach for Attribute Inference Attacks against Synthetic Datacited by 23
  7. 2024Grounding and Evaluation for Large Language Models: Practical Challenges and Lessons Learned (Survey)cited by 20
  8. 2023Generating Private Synthetic Data with Genetic Algorithmscited by 18
  9. 2024Differentially Private Tabular Data Synthesis using Large Language Modelscited by 16
  10. 2022Confidence-Ranked Reconstruction of Census Microdata from Published Statisticscited by 15
  11. 2023Graphical vs. Deep Generative Models: Measuring the Impact of Differentially Private Mechanisms and Budgets on Utilitycited by 13
  12. 2022dpart: Differentially Private Autoregressive Tabular, a General Framework for Synthetic Data Generationcited by 12
38 more
  1. 2023CuTS: Customizable Tabular Synthetic Data Generationcited by 12
  2. 2024The Elusive Pursuit of Reproducing PATE-GAN: Benchmarking, Auditing, Debuggingcited by 12
  3. 2024Privacy Vulnerabilities in Marginals-based Synthetic Datacited by 10
  4. 2022Noise-Aware Statistical Inference with Differentially Private Synthetic Datacited by 9
  5. 2023Post-processing Private Synthetic Data for Improving Utility on Selected Measurescited by 9
  6. 2024Privacy-preserving data release leveraging optimal transport and particle gradient descentcited by 8
  7. 2024Joint Selection: Adaptively Incorporating Public Information for Private Synthetic Datacited by 8
  8. 2025Synthetic Tabular Data: Methods, Attacks and Defensescited by 8
  9. 2022QuerySnout: Automating the Discovery of Attribute Inference Attacks against Query-Based Systemscited by 6
  10. 2023FLAIM: AIM-based Synthetic Data Generation in the Federated Settingcited by 6
  11. 2024Tabular Data Synthesis with Differential Privacy: A Surveycited by 6
  12. 2021Relaxed Marginal Consistency for Differentially Private Query Answeringcited by 5
  13. 2025Is API Access to LLMs Useful for Generating Private Synthetic Tabular Data?cited by 5
  14. 2023A Simple and Practical Method for Reducing the Disparate Impact of Differential Privacycited by 4
  15. 2024Private Regression via Data-Dependent Sufficient Statistic Perturbationcited by 4
  16. 2025The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Datacited by 4
  17. 2025Improving the Generation and Evaluation of Synthetic Data for Downstream Medical Causal Inferencecited by 4
  18. 2022The Tensor Data Platform: Towards an AI-centric Database Systemcited by 3
  19. 2024Sharper Bounds for Chebyshev Moment Matching, with Applicationscited by 3
  20. 2021Rejoinder: Gaussian Differential Privacycited by 2
  21. 2022Private Synthetic Data with Hierarchical Structurecited by 2
  22. 2022Answering Private Linear Queries Adaptively using the Common Mechanismcited by 2
  23. 2023ResidualPlanner+: a scalable matrix mechanism for marginals and beyondcited by 2
  24. 2024A Bias-Variance Decomposition for Ensembles over Multiple Synthetic Datasetscited by 2
  25. 2024Differentially Private Synthetic Data Generation for Relational Databasescited by 2
  26. 2024Privacy without Noisy Gradients: Slicing Mechanism for Generative Model Trainingcited by 2
  27. 2025dpmm: Differentially Private Marginal Models, a Library for Synthetic Tabular Data Generationcited by 2
  28. 2025Private Evolution Convergescited by 2
  29. 2023Private Synthetic Data Meets Ensemble Learningcited by 1
  30. 2024Efficient and Private Marginal Reconstruction with Local Non-Negativitycited by 1
  31. 2026Rethinking Anonymity Claims in Synthetic Data Generation: A Model-Centric Privacy Attack Perspectivecited by 1
  32. 2026Accurate and Scalable Matrix Mechanisms via Divide and Conquercited by 1
  33. 2023PRIMO: Private Regression in Multiple Outcomescited by 0
  34. 2023Partition-based differentially private synthetic data generationcited by 0
  35. 2024Differentially Private Synthetic High-dimensional Tabular Streamcited by 0
  36. 2025Differentially Private Synthetic Data Release for Topics API Outputscited by 0
  37. 2025Privacy Blur: Quantifying Privacy and Utility for Image Data Releasecited by 0
  38. 2026Private Adaptive Covariance Estimation via Gaussian Graphical Modelscited by 0