Preprint · 2023 · Open access

Differentially Private Diffusion Models Generate Useful Synthetic Images

Sahra Ghalebikesabi, Leonard Berrada, Sven Gowal, Sofia Ira Ktena, Robert Stanforth, Jamie Hayes, Soham De, Samuel L. Smith, Olivia Wiles, Borja Balle

arXiv (Cornell University) · Published 27 February 2023 · DOI 10.48550/arxiv.2302.13861

Open PDFView at the publisherCite this paper

In brief

“Differentially Private Diffusion Models Generate Useful Synthetic Images” is a 2023 preprint by Sahra Ghalebikesabi and 9 co-authors, published in arXiv (Cornell University). OpenAlex counts 144 citations as of 19 November 2026. It is open access under a CC BY licence: the full text is free to read at arXiv (Cornell University).

Citations
144
Most cited in
2025 (48)
References
79
Authors
10

Abstract

The ability to generate privacy-preserving synthetic versions of sensitive image datasets could unlock numerous ML applications currently constrained by data availability. Due to their astonishing image generation quality, diffusion models are a prime candidate for generating high-quality synthetic data. However, recent studies have found that, by default, the outputs of some diffusion models do not preserve training data privacy. By privately fine-tuning ImageNet pre-trained diffusion models with more than 80M parameters, we obtain SOTA results on CIFAR-10 and Camelyon17 in terms of both FID and the accuracy of downstream classifiers trained on synthetic data. We decrease the SOTA FID on CIFAR-10 from 26.2 to 9.8, and increase the accuracy from 51.0% to 88.0%. On synthetic data from Camelyon17, we achieve a downstream accuracy of 91.1% which is close to the SOTA of 96.5% when training on the real data. We leverage the ability of generative models to create infinite amounts of data to maximise the downstream prediction performance, and further show how to use synthetic data for hyperparameter tuning. Our results demonstrate that diffusion models fine-tuned with differential privacy can produce useful and provably private synthetic data, even in applications with significant distribution shift between the pre-training and fine-tuning distributions.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Ghalebikesabi, S., Berrada, L., Gowal, S., Ktena, S. I., Stanforth, R., Hayes, J., De, S., Smith, S. L., Wiles, O., & Balle, B. (2023). Differentially Private Diffusion Models Generate Useful Synthetic Images. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2302.13861

MLA 9

Ghalebikesabi, Sahra, et al. “Differentially Private Diffusion Models Generate Useful Synthetic Images.” arXiv (Cornell University), 2023. https://doi.org/10.48550/arxiv.2302.13861.

Chicago

Ghalebikesabi, Sahra, Leonard Berrada, Sven Gowal, Sofia Ira Ktena, Robert Stanforth, Jamie Hayes, Soham De, Samuel L. Smith, Olivia Wiles, and Borja Balle. 2023. “Differentially Private Diffusion Models Generate Useful Synthetic Images.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2302.13861.

Harvard

Ghalebikesabi, S. et al. (2023) ‘Differentially Private Diffusion Models Generate Useful Synthetic Images’, arXiv (Cornell University). doi:10.48550/arxiv.2302.13861.

IEEE

S. Ghalebikesabi et al., “Differentially Private Diffusion Models Generate Useful Synthetic Images,” arXiv (Cornell University), 2023, doi: 10.48550/arxiv.2302.13861.

BibTeX

@misc{ghalebikesabi2023differentially,
  title     = {{Differentially Private Diffusion Models Generate Useful Synthetic Images}},
  author    = {Ghalebikesabi, Sahra and Berrada, Leonard and Gowal, Sven and Ktena, Sofia Ira and Stanforth, Robert and Hayes, Jamie and De, Soham and Smith, Samuel L. and Wiles, Olivia and Balle, Borja},
  howpublished = {arXiv (Cornell University)},
  year      = {2023},
  doi       = {10.48550/arxiv.2302.13861},
  url       = {https://doi.org/10.48550/arxiv.2302.13861}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Citations by year

YearCitationsShare of the best year
202211
20231818
20244141
20254848
20263636

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data

  • differential privacy
  • diffusion models
  • synthetic image generation
  • privacy-preserving data generation
  • FID
  • Camelyon17
  • CIFAR-10
  • hyperparameter optimization

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Ghalebikesabi, S., Berrada, L., Gowal, S., Ktena, S. I., Stanforth, R., Hayes, J., De, S., Smith, S. L., Wiles, O., & Balle, B. (2023). Differentially Private Diffusion Models Generate Useful Synthetic Images. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2302.13861

How many times has this paper been cited?

144 times, according to OpenAlex on 19 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at arXiv (Cornell University): sign in and open the PDF from this page.

Who wrote this paper?

Sahra Ghalebikesabi, Leonard Berrada, Sven Gowal, Sofia Ira Ktena, Robert Stanforth, Jamie Hayes, Soham De, Samuel L. Smith, Olivia Wiles, Borja Balle.

What is the DOI of this paper?

10.48550/arxiv.2302.13861, permanently at https://doi.org/10.48550/arxiv.2302.13861. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 19 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.