Research map / Federated learning and privacy / Differential privacy
Personalized Federated Training of Diffusion Models with Privacy Guarantees
Kumar Kshitij Patel, Bingqing Jiang, A F M Mahfuzul Kabir, Weitong Zhang, Difan Zou, Lingxiao Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use PFDM (shared noisy-data model + personalized local model) instead of non-collaborative training | PFDM · CelebA | FID (minority): 41.38 -> 28.09 | Improved |
| use PFDM instead of non-collaborative training | PFDM · CIFAR-10 | FID (minority): 36.44 -> 35.78 | No clear effect |
| use PFDM instead of non-collaborative training | PFDM · Colorized MNIST | FID (avg): 4.09 -> 3.26 | Improved |
| use PFDM instead of non-private federated training | PFDM · CelebA | FID (avg): 13.15 -> 23.10 | Hurt |
| use PFDM instead of federated DPDM | PFDM · MNIST | FID (avg): 33.73 -> 6.96 | Improved |
| increase number of clients from 4 to 128 with fixed total data (ablation) | PFDM · CIFAR-10 | FID | Improved |
| train global model on noisy clipped data with DP guarantee | PFDM global model · CIFAR-10 membership inference | AUC: 82.13 -> 50.01 | Improved |
| use PFDM global model for reconstruction attack vs no attack | PFDM · CelebA | Recovery score: 1.13 -> 1.12 | No clear effect |
Closest papers by meaning
- A Bayesian Approach for Personalized Federated Learning in Heterogeneous Settings (NeurIPS 2024 · similarity 0.89)
- Leveraging Randomness in Model and Data Partitioning for Privacy Amplification (ICML 2025 · similarity 0.88)
- Federated Learning via Input-Output Collaborative Distillation (AAAI 2024 · similarity 0.88)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.88)
- Private Model Personalization Revisited (ICML 2025 · similarity 0.88)
- PRISM: Privacy-Preserving Improved Stochastic Masking for Federated Generative Models (ICLR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
