Research map / Federated learning and privacy / Communication-efficient FL
Federated Learning Based on Dynamic Regularization
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 |
|---|---|---|---|
| Use dynamic local regularization in FedDyn instead of SCAFFOLD | Federated learning on CIFAR-10, moderate devices, IID · CIFAR-10, IID, target accuracy 84.5 | models transmitted: 637 vs. 1852(2.9$\times$) | Helped |
| Use dynamic local regularization in FedDyn instead of SCAFFOLD | Federated learning on CIFAR-100, moderate devices, Dirichlet (.3) · CIFAR-100, Dirichlet (.3), target accuracy 42.3 | models transmitted: 170 vs. 330(1.9$\times$) | Helped |
| Use dynamic local regularization in FedDyn instead of SCAFFOLD | Federated learning on massively distributed CIFAR-10 · CIFAR-10, massive devices, IID, target accuracy 80.0 | models transmitted: 840 vs. 4000$+$($>4.8$\times$) | Helped |
| Use dynamic local regularization in FedDyn instead of SCAFFOLD | Federated learning on CIFAR-10 with full participation · CIFAR-10, 100% participation, IID, target accuracy 85.0 | models transmitted: 198 vs. 1860(9.4$\times$) | Helped |
| Use dynamic local regularization in FedDyn instead of FedAvg | Federated learning on MNIST with partial participation · MNIST, 10% participation, IID, target accuracy 97.2 | models transmitted: 31 vs. 49 | Helped |
| Use dynamic local regularization in FedDyn instead of SCAFFOLD | Federated learning on CIFAR-10 with unbalanced, heterogeneous data · CIFAR-10, unbalanced data, Dirichlet (.3), target accuracy 80.1 | models transmitted: 274 vs. 652(2.4$\times$) | Helped |
| Vary FedDyn's regularization parameter alpha (ablation) | FedDyn on CIFAR-10, IID, 100 devices, 10% participation · CIFAR-10, IID, 100 devices, 10% participation | test accuracy | Mixed |
| Use FedSplit All, averaging all device models, instead of FedSplit Act, averaging only active devices (ablation) | FedSplit with partial participation on CIFAR-10 · CIFAR-10, 10% participation | convergence performance | Hurt |
Closest papers by meaning
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.89)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.89)
- Efficient Federated Learning against Heterogeneous and Non-stationary Client Unavailability (NeurIPS 2024 · similarity 0.89)
- Federated Learning under Periodic Client Participation and Heterogeneous Data: A New Communication-Efficient Algorithm and Analysis (NeurIPS 2024 · similarity 0.89)
- Communication-Efficient Federated Learning with Accelerated Client Gradient (CVPR 2024 · similarity 0.89)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 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.