Research map / Federated learning and privacy / Communication-efficient FL
Federated Learning under Arbitrary Communication Patterns
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 |
|---|---|---|---|
| asynchronous round-robin communication (RR(2,1)) instead of fully synchronous communication every 5 rounds | local SGD with asynchronous client communication · MNIST / FASHION-MNIST, mixing rate 1 | test accuracy | No clear effect |
| random per-round client communication (RANDOM(1/5)) instead of round-robin with same expected communication | local SGD with random communication · FASHION-MNIST, mixing rate 1/10 | test accuracy: 74.6% vs 77.5% (RR(2,1)) | Hurt |
| imbalanced communication (client i communicates every i rounds) | local SGD with asynchronous communication · CIFAR-10 ResNet-34, mixing rate 1/2 | test accuracy: 38% vs 30% for FULLSGD(1) | Helped |
| decreasing data overlap across clients (mixing rate 1/2 to 0) for round-robin communication | RR(2,5) local SGD · FASHION-MNIST | test accuracy: 68% at mixing rate 0 vs ~80% range at higher mixing | Hurt |
| lowering communication frequency (RR(2,5) vs RR(2,1)) at high data overlap (ablation) | local SGD with asynchronous communication · MNIST/FASHION-MNIST, mixing rate 1 to 1/10 | test accuracy: e.g. RR(2,5) drops 3% on MNIST and FASHION-MNIST when mixing rate goes 1 to 1/10 | No clear effect |
Closest papers by meaning
- Adaptive Federated Learning with Auto-Tuned Clients (ICLR 2024 · similarity 0.90)
- LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging (NeurIPS 2026 · similarity 0.90)
- Communication-Efficient Federated Learning with Accelerated Client Gradient (CVPR 2024 · similarity 0.89)
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.88)
- Federated Learning under Periodic Client Participation and Heterogeneous Data: A New Communication-Efficient Algorithm and Analysis (NeurIPS 2024 · similarity 0.88)
- Efficient Federated Learning against Heterogeneous and Non-stationary Client Unavailability (NeurIPS 2024 · 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.