Research map / Federated learning and privacy / Communication-efficient FL
Decentralized Federated Averaging
Tao Sun, Dongsheng Li, Bao Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| decentralize FedAvg onto a ring graph (DFedAvgM vs FedAvg), IID data | 2NN · MNIST (IID) | test loss/accuracy vs communication rounds and bits | Improved |
| use DFedAvgM instead of DSGD | 2NN · MNIST (IID) | convergence speed per communication round | Improved |
| decentralize FedAvg onto a ring graph (DFedAvgM vs FedAvg), Non-IID data | 2NN · MNIST (Non-IID) | test accuracy (%): 96.81% -> 85% | Hurt |
| lower the quantization bits of communicated models (ablation) | DFedAvgM · MNIST (IID and Non-IID) | training loss, test accuracy, MIA AUC | No clear effect |
| increase local epochs (ablation) | DFedAvgM · MNIST (IID) | training speed and membership privacy | Mixed |
| increase local epochs (ablation) | DFedAvgM · MNIST (Non-IID) | training speed and membership privacy | No clear effect |
| increase communication precision (ablation) | DFedAvgM · Shakespeare LSTM | test accuracy | Improved |
| increase local epochs (ablation) | DFedAvgM · Shakespeare LSTM | test accuracy | Hurt |
| lower the quantization bits of communicated models (ablation) | DFedAvgM · CIFAR10 ResNet20 (IID) | training loss, test accuracy, MIA AUC | No clear effect |
| increase local epochs from 1 to 2 or 5 (ablation) | DFedAvgM · CIFAR10 ResNet20 (IID) | convergence: 1 -> 2 or 5 | Hurt |
Closest papers by meaning
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.88)
- Efficient Federated Learning against Byzantine Attacks and Data Heterogeneity via Aggregating Normalized Gradients (NeurIPS 2025 · similarity 0.87)
- Debiasing Federated Learning with Correlated Client Participation (ICLR 2025 · similarity 0.87)
- Momentum Benefits Non-iid Federated Learning Simply and Provably (ICLR 2024 · similarity 0.87)
- Widening the Network Mitigates the Impact of Data Heterogeneity on FedAvg (ICML 2025 · similarity 0.87)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · 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.
