Research map / Federated learning and privacy / Differential privacy
FedMeNF: Privacy-Preserving Federated Meta-Learning for Neural Fields
Junhyeog Yun, Minui Hong, Gunhee Kim
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 FedMeNF privacy-preserving loss instead of MAML in FedAvg | FedAvg + meta-learning · Cars (NeRF) | PSNR_p (lower is better): 19.73 -> 12.15 | Improved |
| use FedMeNF instead of MAML | FedAvg · Cars (NeRF) | PSNR: 23.08 -> 24.05 | Improved |
| use FedMeNF instead of MAML | FedAvg · PetFace (image) | PSNR: 27.39 -> 27 | No clear effect |
| use FedMeNF instead of MAML | FedAvg · FaceScape (NeRF) | PSNR: 28.59 -> 27.88 | Hurt |
| use differentially private meta-NSGD instead of MAML | FedAvg · PetFace (image) | PSNR: 27.39 -> 5.15 | Hurt |
| increase regularization coefficient gamma to 0.25, 0.5, 0.75 (ablation) | FedMeNF · Cars (NeRF) | PSNR_p improvement over baseline: 18.25%, 35.17%, 38.42%, and 65.28% | Improved |
| meta-learning initialization (FedMeNF) with test-time optimization vs Local | FedMeNF · Cars (NeRF) | PSNR: 40.4% and 74.64% higher PSNR than Local with and without TTO | Improved |
| use FedMeNF instead of Local training for few-step optimization | FedMeNF · Cars (NeRF) | optimization steps: 39 steps vs 8192 steps, about 210x faster | Improved |
| use FedMeNF instead of MAML against membership inference | FedAvg · Cars (NeRF) | MIA accuracy: 95.83 -> 55.56 | Improved |
| use adaptive gamma instead of fixed gamma (ablation) | FedMeNF · Cars NVS (Dirichlet alpha=1.0) | PSNR: 24.32 -> 20.06 | Hurt |
| use adaptive gamma instead of fixed gamma (ablation) | FedMeNF · Cars NVS (Dirichlet alpha=5.0) | PSNR: 24.05 -> 23.19 | Hurt |
| use FedMeNF with fewer input views (2-view) instead of MAML | FedAvg · Cars NVS (2 input views) | PSNR: 20.23 -> 21.27 | Improved |
Closest papers by meaning
- Clients Collaborate: Flexible Differentially Private Federated Learning with Guaranteed Improvement of Utility-Privacy Trade-off (ICML 2025 · similarity 0.88)
- A New Federated Learning Framework Against Gradient Inversion Attacks (AAAI 2025 · similarity 0.88)
- Federated Learning via Input-Output Collaborative Distillation (AAAI 2024 · similarity 0.87)
- DictPFL: Efficient and Private Federated Learning on Encrypted Gradients (NeurIPS 2025 · similarity 0.86)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.86)
- Little Is Enough: Boosting Privacy by Sharing Only Hard Labels in Federated Semi-Supervised Learning (AAAI 2025 · similarity 0.86)
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.
