Research map / Federated learning and privacy / Communication-efficient FL
AdaBest: Minimizing Client Drift in Federated Learning via Adaptive Bias Estimation
Farshid Varno, Marzie Saghayi, Laya Rafiee Sevyeri, Sharut Gupta, Stan Matwin, Mohammad Havaei
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 |
|---|---|---|---|
| replace FedDyn server update with AdaBest adaptive bias estimation (beta=0.96) | FedDyn · CIFAR100, alpha=0.3, 10% client participation | top-1 test accuracy (%): 48.92 -> 54.51 | Improved |
| replace FedDyn with AdaBest | FedDyn · CIFAR100, IID, 10% client participation | top-1 test accuracy (%): 49.60 -> 55.58 | Improved |
| replace FedDyn with AdaBest | FedDyn · CIFAR100, alpha=0.03, 10% client participation | top-1 test accuracy (%): 44.24 -> 48.56 | Improved |
| replace FedDyn with AdaBest | FedDyn · CIFAR100, alpha=0.3, 1% client participation | top-1 test accuracy (%): 50.53 -> 53.62 | Improved |
| replace FedDyn with AdaBest | FedDyn · CIFAR100, alpha=0.03, 1% client participation | top-1 test accuracy (%): 46.27 -> 47.91 | Improved |
| replace FedDyn with AdaBest | FedDyn · EMNIST-L, alpha=0.03, 1% client participation | top-1 test accuracy (%): 92.42 -> 94.49 | Improved |
| replace SCAFFOLD with AdaBest | SCAFFOLD · CIFAR100, alpha=0.3, 10% client participation | top-1 test accuracy (%): 46.34 -> 54.51 | Improved |
| replace FedAvg with AdaBest | FedAvg · CIFAR100, alpha=0.3, 10% client participation | top-1 test accuracy (%): 38.78 -> 54.51 | Improved |
| AdaBest versus FedDyn at full participation | FedDyn · CIFAR100, alpha=0.03, 100% client participation | top-1 test accuracy (%): 46.00 -> 46.16 | No clear effect |
| AdaBest versus SCAFFOLD at full participation | SCAFFOLD · CIFAR100, alpha=0.03, 100% client participation | top-1 test accuracy (%): 46.51 -> 46.16 | Hurt |
| replace FedDyn with AdaBest at full participation | FedDyn · CIFAR100, IID, 100% client participation | top-1 test accuracy (%): 50.61 -> 51.33 | Improved |
| replace SCAFFOLD with AdaBest at full participation | SCAFFOLD · EMNIST-L, alpha=0.3, 100% client participation | top-1 test accuracy (%): 94.53 -> 94.40 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SCAFFOLD — SCAFFOLD: Stochastic Controlled Averaging for Federated Learning (ICML 2020) · better on 9 of 10 shared results
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 Learning with Sample-level Client Drift Mitigation (AAAI 2025 · similarity 0.94)
- FedAdaVR: Adaptive Variance Reduction for Robust Federated Learning under Limited Client Participation (NeurIPS 2026 · similarity 0.92)
- FedImpro: Measuring and Improving Client Update in Federated Learning (ICLR 2024 · similarity 0.90)
- Communication-Efficient Federated Learning with Accelerated Client Gradient (CVPR 2024 · similarity 0.90)
- Efficient Federated Learning against Heterogeneous and Non-stationary Client Unavailability (NeurIPS 2024 · similarity 0.89)
- Adaptive Federated Learning with Auto-Tuned Clients (ICLR 2024 · similarity 0.89)
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.
