Research map / Federated learning and privacy / Communication-efficient FL
FedExP: Speeding Up Federated Averaging via Extrapolation
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 |
|---|---|---|---|
| Replace constant server step size with adaptive FedExP server step size (extrapolation) | FedAvg server update · EMNIST / CIFAR-10 / CIFAR-100 / CINIC-10 | rounds to target accuracy: 1.42x-2.06x fewer rounds than FedAvg | Helped |
| Add server extrapolation to SCAFFOLD (extrapolated SCAFFOLD) (ablation) | SCAFFOLD on CIFAR-10 · CIFAR-10 | test accuracy performance | Helped |
| Use SCAFFOLD variance reduction (no server extrapolation) | FedAvg-style training · CIFAR-10 / CINIC-10 / CIFAR-100 | rounds to target accuracy | No clear effect |
| Use FedAdagrad coordinate-wise adaptivity instead of FedAvg | FedAvg server update · CIFAR-10 / CIFAR-100 | rounds to target accuracy | No clear effect |
| Average last two iterates of FedExP as final model (ablation) | FedExP · neural network FL experiments | test accuracy stability | Helped |
| Set server step size to larger value when client-update diversity is high (gradient diversity) | FedExP · synthetic overparameterized linear regression | convergence speed | Helped |
Closest papers by meaning
- The Power of Extrapolation in Federated Learning (NeurIPS 2024 · similarity 0.89)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.88)
- A Lightweight Method for Tackling Unknown Participation Statistics in Federated Averaging (ICLR 2024 · similarity 0.88)
- Widening the Network Mitigates the Impact of Data Heterogeneity on FedAvg (ICML 2025 · similarity 0.87)
- Momentum Benefits Non-iid Federated Learning Simply and Provably (ICLR 2024 · similarity 0.87)
- Efficient Adaptive Federated Optimization (NeurIPS 2025 · 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.