Get Started
Home
Topics
Search
Library
Research questionHow can federated learning stay reliable when heterogeneous edge clients are intermittently available and availability changes over time?Federated training aggregates updates from clients that may miss rounds because of resource constraints or other uncertainties. Uneven, changing participation can bias the aggregate and degrade training, while accounting for it can increase memory and computation.
AI
Machine Learning
Statistical Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.Resilience Beyond Stationary Client Unavailability: Unlocking Efficient and Unbiased Federated LearningThe source concerns parameter-server federated learning with heterogeneous, stochastic, non-stationary client unavailability and no prior real-time knowledge of availability. It provides convergence analysis for non-convex objectives, a linear-speedup result in special cases, and numerical experiments on real-world datasets across varied unavailability dynamics, with reported light overhead relative to standard FedAvg.research paper · Sep 4, 2026
Related questions
How can federated learning reject poisoned client updates without server validation data or participant history?How can federated learning protect client updates from inference while resisting Byzantine manipulation?How can federated PDE learning define and measure client heterogeneity across different equations?How can personalized federated learning reduce bidirectional model communication under bandwidth constraints without sacrificing accuracy?