mlsanityData to 14 Oct 2026

Large-Scale Public Data Improves Differentially Private Image Generation Quality

arXiv 2309.00008 ↗First posted 4 Aug 2023cs.CV cs.CR cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Public data has been frequently used to improve the privacy-accuracy trade-off of differentially private machine learning, but prior work largely assumes that this data come from the same distribution as the private. In this work, we look at how to use generic large-scale public data to improve the quality of differentially private image generation in Generative Adversarial Networks (GANs), and provide an improved method that uses public data effectively. Our method works under the assumption that the support of the public data distribution contains the support of the private; an example of this is when the public data come from a general-purpose internet-scale image source, while the private data consist of images of a specific type. Detailed evaluations show that our method achieves SOTA in terms of FID score and other metrics compared with existing methods that use public data, and can generate high-quality, photo-realistic images in a differentially private manner.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 35 of its 52 references that are in the map. First posted on arXiv 4 Aug 2023.

Builds on

35 of 52 references.

References · 35

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  4. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  5. 2022Generative Adversarial Networkscited by 5,627
  6. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,101
  7. 2016Deep Learning with Differential Privacycited by 2,728
  8. 2013Describing Textures in the Wildcited by 2,125
  9. 2021Large Language Models Can Be Strong Differentially Private Learnerscited by 412
  10. 2018Assessing Generative Models via Precision and Recallcited by 406
  11. 2021Differentially Private Fine-tuning of Language Modelscited by 400
  12. 2021Opacus: User-Friendly Differential Privacy Library in PyTorchcited by 383
23 more
  1. 2016Semi-supervised Knowledge Transfer for Deep Learning from Private Training Datacited by 348
  2. 2018Scalable Private Learning with PATEcited by 269
  3. 2018Differentially Private Generative Adversarial Networkcited by 235
  4. 2020Differentially Private Learning Needs Better Features (or Much More Data)cited by 235
  5. 2022Unlocking High-Accuracy Differentially Private Image Classification through Scalecited by 224
  6. 2019R'enyi Differential Privacy of the Sampled Gaussian Mechanismcited by 204
  7. 2018Transferring GANs: generating images from limited datacited by 134
  8. 2020Freeze the Discriminator: a Simple Baseline for Fine-Tuning GANscited by 132
  9. 2021Instance-Conditioned GANcited by 121
  10. 2020GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generatorscited by 117
  11. 2019MineGAN: effective knowledge transfer from GANs to target domains with few imagescited by 106
  12. 2022NICO++: Towards Better Benchmarking for Domain Generalizationcited by 86
  13. 2020DP-MERF: Differentially Private Mean Embeddings with Random Features for Practical Privacy-Preserving Data Generationcited by 78
  14. 2018On GANs and GMMscited by 62
  15. 2021Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergencecited by 54
  16. 2019G-PATE: Scalable Differentially Private Data Generator via Private Aggregation of Teacher Discriminatorscited by 53
  17. 2022Bounding Training Data Reconstruction in Private (Deep) Learningcited by 46
  18. 2015Differentially Private Ordinary Least Squarescited by 36
  19. 2021Efficient Conditional GAN Transfer with Knowledge Propagation across Classescited by 13
  20. 2020On Leveraging Pretrained GANs for Generation with Limited Datacited by 11
  21. 2022Improved Input Reprogramming for GAN Conditioningcited by 7
  22. 2022Differentially Private Generative Adversarial Networks with Model Inversioncited by 7
  23. 2021Transferring Unconditional to Conditional GANs with Hyper-Modulationcited by 4

Built on by

Not cited yet.