Research map / Federated learning and privacy / Personalized and heterogeneous FL
Personalized Federated Learning with First Order Model Optimization
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 client-specific weighted combinations of downloaded models instead of one shared FedAvg model | FedFomo · MNIST and CIFAR-10, pathological non-IID, 100 clients | accuracy: 78.83 \pm 9.68 -> 98.81 \pm 1.26 (MNIST); 53.08 \pm 7.40 -> 92.67 \pm 4.21 (CIFAR-10) | Helped |
| Use client-specific weighted combinations of downloaded models instead of one shared FedAvg model | FedFomo · CIFAR-10, pathological non-IID, 15 clients | accuracy: 57.12 \pm 1.14 -> 93.01 \pm 0.96 | No clear effect |
| Use client-specific weighted combinations of downloaded models instead of one shared FedAvg model | FedFomo · MNIST, pathological non-IID, 15 clients | accuracy: 91.97 \pm 2.17 -> 99.63 \pm 0.07 | Mixed |
| Use client-specific weighted combinations of downloaded models instead of one shared FedAvg model | FedFomo · CIFAR-10, pathological non-IID, 15 clients | accuracy: 57.12 \pm 1.14 -> 93.01 \pm 0.96 | Mixed |
| Increase the number of clients from 15 to 100 | FedFomo · Pathological non-IID MNIST and CIFAR-10 | accuracy: 99.62 \pm 2.91 -> 98.81 \pm 1.26 (MNIST n=5); 93.01 \pm 0.96 -> 92.10 \pm 5.20 (CIFAR-10 n=5) | Mixed |
| Add client-specific model evaluation on a validation split reflecting the target distribution | FedFomo · Out-of-local-distribution CIFAR-10 evaluation | accuracy | Helped |
| Use FedProx's proximal term to reduce model-weight divergence | FedProx · Latent-distribution non-IID image classification | accuracy | No clear effect |
| Share 5% of training data across clients | FedAvg · Latent-distribution non-IID image classification | accuracy | Helped |
| Add FedFomo personalization to standard federated training | FedFomo · Pathological non-IID CIFAR-10, evaluation over communication rounds | accuracy | Helped |
Closest papers by meaning
- Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax Guarantees (ICML 2025 · similarity 0.90)
- FedAS: Bridging Inconsistency in Personalized Federated Learning (CVPR 2024 · similarity 0.90)
- pFedClub: Controllable Heterogeneous Model Aggregation for Personalized Federated Learning (NeurIPS 2024 · similarity 0.90)
- Fine-Tuning Personalization in Federated Learning to Mitigate Adversarial Clients (NeurIPS 2024 · similarity 0.90)
- Personalized Federated Learning with Mixture of Models for Adaptive Prediction and Model Fine-Tuning (NeurIPS 2024 · similarity 0.89)
- Decentralized Directed Collaboration for Personalized Federated Learning (CVPR 2024 · 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.