Pith. sign in

REVIEW 5 cited by

Differentially Private Diffusion Models Generate Useful Synthetic Images

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2302.13861 v1 pith:RFRXM2NF submitted 2023-02-27 cs.LG cs.CRcs.CVstat.ML

Differentially Private Diffusion Models Generate Useful Synthetic Images

classification cs.LG cs.CRcs.CVstat.ML
keywords datasyntheticmodelsdiffusionaccuracydownstreamsotaability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original 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.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Privacy Leakage via Output Label Space and Differentially Private Continual Learning

    cs.LG 2024-11 unverdicted novelty 7.0

    Identifies output label space as a privacy side-channel in DP continual learning, formalizes DP for CL, and demonstrates two mitigation methods yielding higher accuracy than prior work.

  2. WHERE to Generate Matters: Budget-Aware Synthetic Augmentation for Label Skewed Federated Learning

    cs.LG 2026-07 conditional novelty 6.0

    Entropy-adaptive per-class budgets let clients generate far fewer synthetic samples yet still close most of the accuracy gap caused by label skew in federated learning.

  3. CollaFuse: Collaborative Diffusion Models

    cs.LG 2024-06 unverdicted novelty 6.0

    CollaFuse enables collaborative diffusion model training by splitting computation between resource-limited clients and a central server to reduce local burden and raw data sharing.

  4. PeerCheck: Enhancing LLM-Generated Academic Reviews Towards Human-Level Quality

    cs.CL 2026-06 unverdicted novelty 5.0

    PeerCheck finds that chain-of-thought prompting improves LLM academic reviews while retrieval-augmented generation sometimes lowers quality, and that LLMs and humans emphasize different aspects of papers.

  5. DP-SAPF: Saliency-Aware Parameter Fine-tuning of Public Models for Differentially Private Image Synthesis

    cs.CR 2026-05 unverdicted novelty 5.0

    DP-SAPF uses gradient-magnitude saliency after noise addition to select LoRA targets for DP fine-tuning of public models, reporting improved synthetic image utility and fidelity on four datasets with lower compute.