Private Image Generation with Dual-Purpose Auxiliary Classifier
Chen Chen, Daochang Liu, Siqi Ma, Surya Nepal, Chang Xu
Abstract
Privacy-preserving image generation has been important for segments such as medical domains that have sensitive and limited data. The benefits of guaranteed privacy come at the costs of generated images' quality and utility due to the privacy budget constraints. The utility is currently measured by the gen2real accuracy (g2r%), i.e., the accuracy on real data of a downstream classifier trained using generated data. However, apart from this standard utility, we identify the "reversed utility" as another crucial aspect, which computes the accuracy on generated data of a classifier trained using real data, dubbed as real2gen accuracy (r2g%). Jointly considering these two views of utility, the standard and the reversed, could help the generation model better improve transferability between fake and real data. Therefore, we propose a novel private image generation method that incorporates a dual-purpose auxiliary classifier, which alternates between learning from real data and fake data, into the training of differentially private GANs. Additionally, our deliberate training strategies such as sequential training contributes to accelerating the generator's convergence and further boosting the performance upon exhausting the privacy budget. Our results achieve new state-of-the-arts over all metrics on three benchmarks: MNIST, Fashion-MNIST, and CelebA.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
2 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Membership Inference Attacks Against Machine Learning Models cites Deep Learning with Differential Privacy
- DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation cites Membership Inference Attacks Against Machine Learning Models
- DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation cites GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators
- DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation cites Deep Learning with Differential Privacy
- Towards Memorization-Free Diffusion Models cites Deep Learning with Differential Privacy
- Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergence cites DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation
- Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergence cites GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators
- Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergence cites Deep Learning with Differential Privacy
- Private Gradient Estimation is Useful for Generative Modeling cites DataLens: Scalable Privacy Preserving Training via Gradient Compression and Aggregation
- Private Gradient Estimation is Useful for Generative Modeling cites GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators
- Private Gradient Estimation is Useful for Generative Modeling cites Deep Learning with Differential Privacy
- GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private Generators cites Deep Learning with Differential Privacy
Cited by top-tier papers2
Builds on6
- Deep Learning with Differential PrivacyMartín Abadi, Andy Chu, Ian J. Goodfellow, H. Brendan McMahan et al.CCS 2016 · 7,620 citations
- Membership Inference Attacks Against Machine Learning ModelsReza Shokri, Marco Stronati, Congzheng Song, Vitaly ShmatikovS&P 2017 · 5,137 citations
- GS-WGAN: A Gradient-Sanitized Approach for Learning Differentially Private GeneratorsDingfan Chen, Tribhuvanesh Orekondy, Mario FritzNeurIPS 2020 · 228 citations
- Don't Generate Me: Training Differentially Private Generative Models with Sinkhorn DivergenceTianshi Cao, Alex Bie, Arash Vahdat, Sanja Fidler et al.NeurIPS 2021 · 88 citations
- DataLens: Scalable Privacy Preserving Training via Gradient Compression and AggregationBoxin Wang, Fan Wu, Yunhui Long, Luka Rimanic et al.CCS 2021 · 45 citations
Related papers
- Conditional GANs with Auxiliary Discriminative ClassifierLiang Hou, Qi Cao, Huawei Shen, Siyuan Pan et al.ICML 2022 · 49 citations
- Variational Model Inversion AttacksKuan-Chieh Wang, Yan Fu, Ke Li, Ashish Khisti et al.NeurIPS 2021 · 142 citations
- A Novel Confidence Guided Training Method for Conditional GANs with Auxiliary ClassifierQi Chen, Wenjie Liu, Hu DingACM MM 2024
- Differentially Private Image Classification by Learning Priors from Random ProcessesXinyu Tang, Ashwinee Panda, Vikash Sehwag, Prateek MittalNeurIPS 2023 · 34 citations
- SoK: Can Synthetic Images Replace Real Data? A Survey of Utility and Privacy of Synthetic Image GenerationYunsung Chung, Yunbei Zhang, Nassir Marrouche, Jihun HammUSENIX Security 2025
