Research map / Federated learning and privacy / Personalized and heterogeneous FL
Personalized Federated Learning with Theoretical Guarantees: A Model-Agnostic Meta-Learning Approach
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 |
|---|---|---|---|
| Per-FedAvg (HF) meta-learned init instead of FedAvg + one-step update | FedAvg · MNIST tau=10 | test accuracy: 75.96% -> 79.85% | Helped |
| Per-FedAvg (HF) instead of FedAvg + update | FedAvg · CIFAR-10 tau=4 alpha=0.01 | test accuracy: 35.97% -> 46.32% | Helped |
| Per-FedAvg (HF) instead of FedAvg + update under stronger heterogeneity | FedAvg · CIFAR-10 diff. hetero. | test accuracy: 58.59% -> 71.25% | Helped |
| Per-FedAvg (FO) first-order approximation instead of FedAvg + update | FedAvg · CIFAR-10 tau=4 | test accuracy: 38.38% -> 34.04% (alpha=0.001); 40.49% -> 46.98% (tau=10) | Mixed |
| Hessian-free approximation (HF) instead of first-order (FO) approximation (ablation) | Per-FedAvg · CIFAR-10 tau=4 alpha=0.01 | test accuracy: 25.32% -> 46.32% | Helped |
| increase test-time step alpha 0.001 -> 0.01 (ablation) | Per-FedAvg (HF) · CIFAR-10 tau=4 | test accuracy: 43.73% -> 46.32% | Helped |
| increase test-time step alpha 0.001 -> 0.01 (ablation) | Per-FedAvg (FO) · CIFAR-10 tau=4 | test accuracy: 34.04% -> 25.32% | Hurt |
| reduce local steps tau 10 -> 4 (ablation) | Per-FedAvg (HF) · CIFAR-10 alpha=0.001 | test accuracy: 50.44% -> 43.73% | Hurt |
Closest papers by meaning
- Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax Guarantees (ICML 2025 · similarity 0.91)
- Decentralized Directed Collaboration for Personalized Federated Learning (CVPR 2024 · similarity 0.90)
- pFedClub: Controllable Heterogeneous Model Aggregation for Personalized Federated Learning (NeurIPS 2024 · similarity 0.90)
- Adaptive Latent-Space Constraints in Personalized Federated Learning (NeurIPS 2025 · similarity 0.89)
- FedAS: Bridging Inconsistency in Personalized Federated Learning (CVPR 2024 · similarity 0.89)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.89)
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.