Client-centric Federated Learning
Primary Area: general machine learning (i.e., none of the above)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: Federated Learning, Client-Centric, Client Autonomy and Asynchrony, Distribution Estimation
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
TL;DR: We propose a novel client-centric federated learning paradigm, which emphasizes on the autonomy, efficiency, and effectiveness of clients' model training.
Abstract: Conventional federated learning (FL) frameworks follow a server-centric model where the server determines session initiation and client participation. We introduce Client-Centric Federated Learning (CCFL), a novel client-centric FL framework that puts clients as the driving role of FL sessions. In CCFL, each client independently and asynchronously updates its model by uploading a locally trained model to the server and receiving a customized model tailored to its local task. The server maintains a repository of cluster models, iteratively refining them using received client models. Our framework accommodates complex dynamics in clients' data distributions, characterized by time-varying mixtures of cluster distributions, enabling rapid adaptation to new tasks with high performance. We propose novel strategies for accurate server estimation of clients' data distributions. CCFL offers clients complete autonomy for model updates, enhances model accuracy, and significantly reduces client computation, communication, and waiting time. We provide a theoretical analysis of CCFL's convergence. Extensive experiments across various datasets and system settings highlight CCFL's substantial advantages in model performance and computation efficiency over baselines.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3372
Loading