Research map / Federated learning and privacy / Communication-efficient FL
A Unified Analysis of Federated Learning with Arbitrary Client Participation
Shiqiang Wang, Mingyue Ji
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 update amplification (eta=10, P=500) to FedAvg (ablation) | Generalized FedAvg (Algorithm 1) · FashionMNIST and CIFAR-10, periodic client availability | test performance (plotted) | Improved |
| replace amplified Algorithm 1 with wait-for-all using minibatch gradients (ablation) | Algorithm 1 with amplification · FashionMNIST and CIFAR-10 | test performance (plotted) | Hurt |
| use full-dataset gradients (wait-full) instead of minibatch gradients in the wait-for-all method (ablation) | wait-for-all baseline · FashionMNIST and CIFAR-10 | test performance (plotted) | No clear effect |
| use amplification interval P=1 instead of P=500 with eta=10 (ablation) | Algorithm 1 with amplification · FashionMNIST and CIFAR-10 | test performance (plotted) | Hurt |
| use P=100 or P=300 instead of P=500 (matching participation cycle) (ablation) | Algorithm 1 with amplification, eta=10 · FashionMNIST and CIFAR-10 | test performance (plotted) | No clear effect |
| increase P to 700 or 900 beyond the participation cycle of 500 (ablation) | Algorithm 1 with amplification, eta=10 · CIFAR-10 | test performance (plotted) | Hurt |
| use regularized participation (equal averaged weights over P rounds) in theory | Generalized FedAvg convergence bound · theory, nonconvex | convergence error bound | Improved |
Closest papers by meaning
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.95)
- Federated Learning under Periodic Client Participation and Heterogeneous Data: A New Communication-Efficient Algorithm and Analysis (NeurIPS 2024 · similarity 0.94)
- Understanding Server-Assisted Federated Learning in the Presence of Incomplete Client Participation (ICML 2024 · similarity 0.91)
- A Lightweight Method for Tackling Unknown Participation Statistics in Federated Averaging (ICLR 2024 · similarity 0.91)
- Debiasing Federated Learning with Correlated Client Participation (ICLR 2025 · similarity 0.91)
- Efficient Federated Learning against Heterogeneous and Non-stationary Client Unavailability (NeurIPS 2024 · similarity 0.90)
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.
