Research map / Federated learning and privacy / Communication-efficient FL
FedNAR: Federated Optimization with Normalized Annealing Regularization
Junbo Li, Ang Li, Chong Tian, Qirong Ho, Eric Xing, Hongyi 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 |
|---|---|---|---|
| Add FedNAR to FedAvg | FedAvg · CIFAR-10 (α = 0.3) | test accuracy: 85.19 -> 87.64 | Improved |
| Add FedNAR to FedAvg | FedAvg · CIFAR-10 (α = 1) | test accuracy: 88.50 -> 89.55 | Improved |
| Add FedNAR to FedAvg | FedAvg · CIFAR-10 (α = 10) | test accuracy: 89.45 -> 90.23 | Improved |
| Add FedNAR to FedProx | FedProx · CIFAR-10 (α = 0.3) | test accuracy: 85.02 -> 87.45 | Improved |
| Add FedNAR to FedProx | FedProx · CIFAR-10 (α = 1) | test accuracy: 88.56 -> 89.58 | Improved |
| Add FedNAR to FedProx | FedProx · CIFAR-10 (α = 10) | test accuracy: 89.37 -> 90.38 | Improved |
| Add FedNAR to SCAFFOLD | SCAFFOLD · CIFAR-10 (α = 0.3, 1, 10) | test accuracy: 86.89 -> 89.17; 89.52 -> 91.13; 90.64 -> 91.85 | Improved |
| Add FedNAR to FedExP | FedExP · CIFAR-10 (α = 0.3, 1, 10) | test accuracy: 86.46 -> 88.55; 88.28 -> 89.51; 88.82 -> 89.69 | Improved |
| Add FedNAR to FedAdam | FedAdam · CIFAR-10 (α = 0.3, 1, 10) | test accuracy: 84.12 -> 85.81; 87.35 -> 87.78; 88.86 -> 89.56 | Improved |
| Add FedNAR to FedAvgm | FedAvgm · CIFAR-10 (α = 0.3, 1, 10) | test accuracy: 90.33 -> 90.43; 91.60 -> 91.76; 92.19 -> 91.15 | Mixed |
| Add FedNAR to FedAvg | FedAvg · Shakespeare (WD = 10^{-4}, 10^{-3}) | accuracy: 42.44 -> 44.37; 40.05 -> 44.69 | Improved |
| Add FedNAR to FedProx | FedProx · Shakespeare (WD = 10^{-4}, 10^{-3}) | accuracy: 41.92 -> 43.57; 40.91 -> 40.26 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- SCAFFOLD — SCAFFOLD: Stochastic Controlled Averaging for Federated Learning (ICML 2020) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Federated Optimization with Doubly Regularized Drift Correction (ICML 2024 · similarity 0.88)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.87)
- Why Do We Need Weight Decay in Modern Deep Learning? (NeurIPS 2024 · similarity 0.87)
- Improving Generalization in Federated Learning with Highly Heterogeneous Data via Momentum-Based Stochastic Controlled Weight Averaging (ICML 2025 · similarity 0.87)
- Locally Estimated Global Perturbations are Better than Local Perturbations for Federated Sharpness-aware Minimization (ICML 2024 · similarity 0.87)
- FedHyper: A Universal and Robust Learning Rate Scheduler for Federated Learning with Hypergradient Descent (ICLR 2024 · 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.
