Taming Client Availability in Federated Learning in the Presence of Arbitrary and Unknown Dynamics
Year: 2025
Type: article
Abstract: Federated learning (FL) is a prominent distributed learning framework that allows clients to train machine learning models orchestrated by a parameter server (PS) [3]. Unfortunately, its practical implementation is fundamentally hindered by heterogeneous local data distribution and intermittent client availability [5]. Moreover, the availability of a client may be highly dynamic and hard to predict.
Author: Ming Xiang
Institution: Northeastern University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
Awards: G2169926672, Collaborative Research: CNS Core: Medium: Data-Centric Networks for Distributed Learning, G5259331294, Collaborative Research: CNS Core: Medium: Data-Centric Networks for Distributed Learning, G7452299184, CAREER: Strengthening the Theoretical Foundations of Federated Learning: Utilizing Underlying Data Statistics in Mitigating Heterogeneity and Client Faults, G981292923
OpenAlex ID: W4408556794