Research map / Federated learning and privacy / Personalized and heterogeneous FL
Personalized Federated Learning using Hypernetworks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a shared hypernetwork to generate personalized client models | Personalized federated learning · CIFAR10, CIFAR100, and Omniglot heterogeneous-client tests | test accuracy | Helped |
| Add a personalized classifier per client (ablation) | pFedHN · Omniglot, 50 clients | test accuracy: 72.03 \pm 1.08 -> 81.89 \pm 0.15 | Helped |
| Add a personalized classifier per client (ablation) | pFedHN · CIFAR10, heterogeneous, 10 clients | test accuracy: 90.83\pm 1.56 -> 92.47 \pm 1.63 | Helped |
| Add a personalized classifier per client (ablation) | pFedHN · CIFAR100, heterogeneous, 100 clients | test accuracy: \mathbf{53.24 \pm 0.31} -> 52.4 \pm 0.74 | Hurt |
| Train using multiple local optimization steps and send the model update direction (ablation) | pFedHN · CIFAR10, 50 clients | test accuracy | Helped |
| Vary the number of local optimization steps (ablation) | pFedHN · CIFAR10, 50 clients | test accuracy | No clear effect |
| Vary client embedding dimension across 5, 15, 25, and 35 (ablation) | pFedHN · CIFAR10, 50 clients | test accuracy | No clear effect |
| Increase hypernetwork depth from 3 to 4 hidden layers (ablation) | pFedHN · CIFAR10, 50 clients | test accuracy: 88.38 -> 88.42 | No clear effect |
| Apply spectral normalization to all hypernetwork layers (ablation) | pFedHN · CIFAR10, 10 clients | test accuracy: 90.83\pm 1.56 -> 90.94 \pm 2.18 | No clear effect |
| Use fixed class-proportion client embeddings instead of trainable embeddings (ablation) | pFedHN · CIFAR10, 50 clients, novel-client setup | performance: 84.12\pm0.42 and 83.92\pm0.36 | No clear effect |
Closest papers by meaning
- PeFLL: Personalized Federated Learning by Learning to Learn (ICLR 2024 · similarity 0.91)
- pFedClub: Controllable Heterogeneous Model Aggregation for Personalized Federated Learning (NeurIPS 2024 · similarity 0.90)
- PFedCS: A Personalized Federated Learning Method for Enhancing Collaboration among Similar Classifiers (AAAI 2025 · similarity 0.89)
- FedAS: Bridging Inconsistency in Personalized Federated Learning (CVPR 2024 · similarity 0.88)
- Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax Guarantees (ICML 2025 · similarity 0.88)
- PerFedRLNAS: One-for-All Personalized Federated Neural Architecture Search (AAAI 2024 · similarity 0.87)
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.