Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedProto: Federated Prototype Learning across Heterogeneous Clients
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 |
|---|---|---|---|
| exchange and aggregate class prototypes rather than model parameters or gradients | federated learning across heterogeneous clients · MNIST, FEMNIST, CIFAR10 | test average accuracy | Helped |
| communicate class prototypes rather than model parameters | FedProto · MNIST, FEMNIST, CIFAR10 | communicated parameters per round | Helped |
| use heterogeneous client architectures (ablation) | FedProto · FEMNIST | test average accuracy: 96.82$\pm$1.75 -> 97.10$\pm$1.63; 94.93$\pm$1.61 -> 94.83$\pm$1.60; 93.67$\pm$2.23 -> 93.76$\pm$2.30 | Mixed |
| use heterogeneous client architectures (ablation) | FedProto · CIFAR10 | test average accuracy: 84.49$\pm$1.97 -> 83.63$\pm$1.60; 79.12$\pm$2.03 -> 79.49$\pm$1.78; 77.08$\pm$1.98 -> 76.94$\pm$1.33 | Mixed |
| use heterogeneous client architectures (ablation) | FedProto · MNIST | test average accuracy: 97.13$\pm$0.30 -> 97.07$\pm$0.50; 96.80$\pm$0.41 -> 96.65$\pm$0.31; 96.70$\pm$0.29 -> 96.22$\pm$0.36 | Hurt |
| increase the prototype regularization weight λ from 0 to 1 (ablation) | FedProto on FEMNIST · FEMNIST (n=3, k=100) | average test accuracy | Helped |
| increase the prototype regularization weight beyond λ=1 (ablation) | FedProto on FEMNIST · FEMNIST (n=3, k=100) | average test accuracy: 6\% | Hurt |
| reduce the number of training samples available per client | FedProto compared with FedAvg · MNIST | test accuracy | Helped |
Closest papers by meaning
- FedImpro: Measuring and Improving Client Update in Federated Learning (ICLR 2024 · similarity 0.91)
- Taming Cross-Domain Representation Variance in Federated Prototype Learning with Heterogeneous Data Domains (NeurIPS 2024 · similarity 0.89)
- FedSA: A Unified Representation Learning via Semantic Anchors for Prototype-based Federated Learning (AAAI 2025 · similarity 0.89)
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.89)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.89)
- Federated Learning with Sample-level Client Drift Mitigation (AAAI 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.