Research map / Federated learning and privacy / Communication-efficient FL
Delayed Gradient Averaging: Tolerate the Communication Latency for Federated Learning
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 |
|---|---|---|---|
| add gradient correction term to delayed gradients (K=5, D=5) (ablation) | DGA · CIFAR-10 | accuracy: 88.7 -> 89.2 | Helped |
| add gradient correction term to delayed gradients (K=5, D=10) (ablation) | DGA · CIFAR-10 | accuracy: 86.9 -> 89.3 | Helped |
| add gradient correction term to delayed gradients (K=5, D=15) (ablation) | DGA · CIFAR-10 | accuracy: 85.5 -> 89.0 | Helped |
| add gradient correction term to delayed gradients (K=5, D=20) (ablation) | DGA · CIFAR-10 | accuracy: 84.2 -> 88.7 | Helped |
| use DGA (K=5, D=20) instead of FedAvg (K=5) at 1s latency | FedAvg · CIFAR-10 i.i.d | speedup / accuracy: 88.7 -> 88.6; 1x -> 3.16 | Helped |
| use DGA (K=5, D=20) instead of FedAvg (K=5) at 1s latency | FedAvg · Shakespeare i.i.d | speedup / accuracy: 47.6 -> 47.1; 1x -> 4.07 | Helped |
| increase local steps K in FedAvg from 5 to 20 to amortize latency | FedAvg · ImageNet non-i.i.d | accuracy: 55.4 -> 48.6 | Hurt |
| increase local steps K in FedAvg from 5 to 20 | FedAvg · CIFAR-10 non-i.i.d | accuracy: 48.2 -> 43.9 | Hurt |
| use DGA (K=5, D=20) instead of FedAvg (K=20) | FedAvg (K=20) · Shakespeare non-i.i.d | accuracy: 30.1 -> 36.3 | Helped |
| use DGA (K=5, D=20) instead of FedAvg (K=20) | FedAvg (K=20) · ImageNet non-i.i.d | accuracy: 48.6 -> 54.9 | Helped |
Closest papers by meaning
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.89)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.87)
- Communication-Efficient Federated Learning with Accelerated Client Gradient (CVPR 2024 · similarity 0.87)
- Federated Optimization with Doubly Regularized Drift Correction (ICML 2024 · similarity 0.87)
- SparsyFed: Sparse Adaptive Federated Learning (ICLR 2025 · similarity 0.87)
- Accelerating Federated Learning with Quick Distributed Mean Estimation (ICML 2024 · similarity 0.86)
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.