mlsanityData to 7 Oct 2026

Differentially Private Synthetic Data via Foundation Model APIs 1: Images

arXiv 2305.15560 ↗First posted 24 May 2023cs.CV cs.CR cs.LGcited by 63

Not on the map. Why?

Abstract (arXiv)

Generating differentially private (DP) synthetic data that closely resembles the original private data is a scalable way to mitigate privacy concerns in the current data-driven world. In contrast to current practices that train customized models for this task, we aim to generate DP Synthetic Data via APIs (DPSDA), where we treat foundation models as blackboxes and only utilize their inference APIs. Such API-based, training-free approaches are easier to deploy as exemplified by the recent surge in the number of API-based apps. These approaches can also leverage the power of large foundation models which are only accessible via their inference APIs. However, this comes with greater challenges due to strictly more restrictive model access and the need to protect privacy from the API provider. In this paper, we present a new framework called Private Evolution (PE) to solve this problem and show its initial promise on synthetic images. Surprisingly, PE can match or even outperform state-of-the-art (SOTA) methods without any model training. For example, on CIFAR10 (with ImageNet as the public data), we achieve FID <= 7.9 with privacy cost ε = 0.67, significantly improving the previous SOTA from ε = 32. We further demonstrate the promise of applying PE on large foundation models such as Stable Diffusion to tackle challenging private datasets with a small number of high-resolution images. The code and data are released at https://github.com/microsoft/DPSDA.

Key facts. Cited by 63 papers in the map (data through 7 Oct 2026). Builds on 46 of its 74 references that are in the map. First posted on arXiv 24 May 2023.

Builds on

46 of 74 references.

References · 46

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  5. 2020Denoising Diffusion Implicit Modelscited by 9,655
  6. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  7. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  8. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,539
  9. 2022Generative Adversarial Networkscited by 5,603
  10. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
  11. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  12. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,736
34 more
  1. 2016Wide Residual Networkscited by 3,057
  2. 2016Deep Learning with Differential Privacycited by 2,682
  3. 2020Extracting Training Data from Large Language Modelscited by 2,144
  4. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  5. 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
  6. 2021Membership Inference Attacks From First Principlescited by 839
  7. 2022Quantifying Memorization Across Neural Language Modelscited by 755
  8. 2023Extracting Training Data from Diffusion Modelscited by 723
  9. 2021Large Language Models Can Be Strong Differentially Private Learnerscited by 405
  10. 2021Differentially Private Fine-tuning of Language Modelscited by 392
  11. 2021Opacus: User-Friendly Differential Privacy Library in PyTorchcited by 375
  12. 2020Label-Only Membership Inference Attackscited by 371
  13. 2022Unlocking High-Accuracy Differentially Private Image Classification through Scalecited by 218
  14. 2018Improving the Gaussian Mechanism for Differential Privacy: Analytical Calibration and Optimal Denoisingcited by 216
  15. 2021Numerical Composition of Differential Privacycited by 174
  16. 2022Reconstructing Training Data from Trained Neural Networkscited by 129
  17. 2022Differentially Private Diffusion Modelscited by 118
  18. 2021Large-Scale Differentially Private BERTcited by 117
  19. 2022Synthetic Text Generation with Differential Privacy: A Simple and Practical Recipecited by 108
  20. 2022Truth Serum: Poisoning Machine Learning Models to Reveal Their Secretscited by 94
  21. 2023Differentially Private Diffusion Models Generate Useful Synthetic Imagescited by 85
  22. 2020DP-MERF: Differentially Private Mean Embeddings with Random Features for Practical Privacy-Preserving Data Generationcited by 78
  23. 2021Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergencecited by 54
  24. 2022Exploring the Limits of Differentially Private Deep Learning with Group-wise Clippingcited by 52
  25. 2022When Does Differentially Private Learning Not Suffer in High Dimensions?cited by 48
  26. 2023Why Is Public Pretraining Necessary for Private Model Training?cited by 36
  27. 2020Differentially Private Clustering: Tight Approximation Ratioscited by 34
  28. 2022Pre-trained Perceptual Features Improve Differentially Private Image Generationcited by 28
  29. 2015Differentially Private k-Means Clusteringcited by 27
  30. 2021Hermite Polynomial Features for Private Data Generationcited by 21
  31. 2022On the Privacy Properties of GAN-generated Samplescited by 16
  32. 2020Differentially Private Set Unioncited by 13
  33. 2023Summary Statistic Privacy in Data Sharingcited by 6
  34. 2022Differentially Private Heatmapscited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 5; 2024: 16; 2025: 23; 2026: 19.

63 papers.

Sort:

By area:

  1. 2023Textbooks Are All You Needcited by 561
  2. 2024Differentially Private Synthetic Data via Foundation Model APIs 2: Textcited by 74
  3. 2024Facilitating Multi-turn Function Calling for LLMs via Compositional Instruction Tuningcited by 40
  4. 2024Privacy-Preserving Instructions for Aligning Large Language Modelscited by 34
  5. 2024On the Challenges and Opportunities in Generative AIcited by 31
  6. 2024PrE-Text: Training Language Models on Private Federated Data in the Age of LLMscited by 29
  7. 2023DP-LDMs: Differentially Private Latent Diffusion Modelscited by 28
  8. 2024Large Language Models for Constructing and Optimizing Machine Learning Workflows: A Surveycited by 19
  9. 2024Differentially Private Fine-Tuning of Diffusion Modelscited by 18
  10. 2024Differentially Private Tabular Data Synthesis using Large Language Modelscited by 16
  11. 2025Synthesizing Privacy-Preserving Text Data via Finetuning without Finetuning Billion-Scale LLMscited by 15
  12. 2025POPri: Private Federated Learning using Preference-Optimized Synthetic Datacited by 14
38 more
  1. 2025Differentially Private Synthetic Data via APIs 3: Using Simulators Instead of Foundation Modelcited by 13
  2. 2023PrivImage: Differentially Private Synthetic Image Generation using Diffusion Models with Semantic-Aware Pretrainingcited by 12
  3. 2025DPImageBench: A Unified Benchmark for Differentially Private Image Synthesiscited by 11
  4. 2024Synth-Empathy: Towards High-Quality Synthetic Empathy Datacited by 10
  5. 2025InvisibleInk: High-Utility and Low-Cost Text Generation with Differential Privacycited by 10
  6. 2024LLM-based Privacy Data Augmentation Guided by Knowledge Distillation with a Distribution Tutor for Medical Text Classificationcited by 8
  7. 2025Synthesize Privacy-Preserving High-Resolution Images via Private Textual Intermediariescited by 8
  8. 2024Efficiently Computing Similarities to Private Datasetscited by 7
  9. 2025Struct-Bench: A Benchmark for Differentially Private Structured Text Generationcited by 7
  10. 2024SynthVLM: Towards High-Quality and Efficient Synthesis of Image-Caption Datasets for Vision-Language Modelscited by 6
  11. 2025Gaussian DP for Reporting Differential Privacy Guarantees in Machine Learningcited by 6
  12. 2025From Easy to Hard: Building a Shortcut for Differentially Private Image Synthesiscited by 6
  13. 2025Is API Access to LLMs Useful for Generating Private Synthetic Tabular Data?cited by 5
  14. 2025Clustering and Median Aggregation Improve Differentially Private Inferencecited by 5
  15. 2023A Unified View of Differentially Private Deep Generative Modelingcited by 4
  16. 2025Generating Synthetic Data with Formal Privacy Guarantees: State of the Art and the Road Aheadcited by 4
  17. 2025Towards Harnessing the Collaborative Power of Large and Small Models for Domain Taskscited by 4
  18. 2023Hot PATE: Private Aggregation of Distributions for Diverse Taskcited by 3
  19. 2024Synthetic Data Aided Federated Learning Using Foundation Modelscited by 3
  20. 2025RewardDS: Privacy-Preserving Fine-Tuning for Large Language Models via Reward Driven Data Synthesiscited by 3
  21. 2026Privasis: Synthesizing the Largest "Public" Private Dataset from Scratchcited by 3
  22. 2024Differentially Private Synthetic Data with Private Density Estimationcited by 2
  23. 2024Differentially Private Kernel Density Estimationcited by 2
  24. 2025Private Text Generation by Seeding Large Language Model Promptscited by 2
  25. 2025Private Evolution Convergescited by 2
  26. 2025ACTG-ARL: Differentially Private Conditional Text Generation with RL-Boosted Controlcited by 2
  27. 2026Differentially Private Synthetic Data via APIs 4: Tabular Datacited by 2
  28. 2025Optimal Differentially Private Sampling of Unbounded Gaussianscited by 1
  29. 2026Paradox of De-identification: A Critique of HIPAA Safe Harbour in the Age of LLMscited by 1
  30. 2026Differentially Private Preference Data Synthesis for Large Language Model Alignmentcited by 1
  31. 2024A Comprehensive Sustainable Framework for Machine Learning and Artificial Intelligencecited by 0
  32. 2025Personalized Federated Training of Diffusion Models with Privacy Guaranteescited by 0
  33. 2025Private Training & Data Generation by Clustering Embeddingscited by 0
  34. 2025Differentially Private Synthetic Data Release for Topics API Outputscited by 0
  35. 2025Latent Zoning Network: A Unified Principle for Generative Modeling, Representation Learning, and Classificationcited by 0
  36. 2025Secret-Protected Evolution for Differentially Private Synthetic Text Generationcited by 0
  37. 2026From Easy to Hard++: Promoting Differentially Private Image Synthesis Through Spatial-Frequency Curriculumcited by 0
  38. 2026Minimax optimal differentially private synthetic data for smooth queriescited by 0

50 most cited of 63 All 63 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025PCEvolve: Private Contrastive Evolution for Synthetic Dataset Generation via Few-Shot Private Data and Generative APIs
  2. 2024Leveraging Programmatically Generated Synthetic Data for Differentially Private Diffusion Training
  3. 2024SafeSynthDP: Leveraging Large Language Models for Privacy-Preserving Synthetic Data Generation Using Differential Privacy
  4. 2025Enabling PSO-Secure Synthetic Data Sharing Using Diversity-Aware Diffusion Models
  5. 2025DP-DocLDM: Differentially Private Document Image Generation using Latent Diffusion Models
  6. 2024DP-TLDM: Differentially Private Tabular Latent Diffusion Model
  7. 2026Private Seeds, Public LLMs: Realistic and Privacy-Preserving Synthetic Data Generation
  8. 2024Efficient Differentially Private Fine-Tuning of Diffusion Models