“Federated Learning via Indirect Communications”
Tuesday, Jan. 9 at 1:00pm
LAR 234
Add to Calendar
Abstract
Federated Learning (FL) represents an evolving paradigm aimed at achieving distributed and privacy-preserving Machine Learning (ML), with applications spanning a diverse range of use cases. While substantial efforts have been invested in refining FL algorithms, the predominant focus has centered on conventional communication scenarios, where clients directly communicate with the parameter server as needed. However, real-world situations often present challenges, leading to sporadic or entirely absent direct client-server interactions. These instances may arise due to disruptions in communication infrastructure or its complete absence. In response to this challenge, we propose a novel framework designed to operate effectively under demanding communication conditions, leveraging indirect client-server communications. This talk will specifically delve into one such strategy: opportunistic mobile relaying. This approach harnesses client mobility to facilitate device-to-device communications among clients, aiming to enhance the frequency of client-server interactions and, consequently, expedite the convergence of FL.
Biography
Dr. Jie Xu is an Associate Professor in the Department of Electrical and Computer Engineering at the University of Miami. He earned his Ph.D. in Electrical Engineering from UCLA in 2015, preceded by his completion of both BS and MS degrees at Tsinghua University in China. Dr. Xu’s research interests are at the intersection of edge computing, machine learning and wireless networking. He focuses on advancing machine learning, optimization, and statistical signal processing algorithms to enhance the performance of current and future computing and networking systems. Additionally, he develops new systems and architectures tailored for emerging applications heavily reliant on machine learning. Dr. Xu has published over 100 research papers in leading journals and conferences, with more than 5000 citations. His research has been supported by federal agencies including National Science Foundation (NSF) and Army Research Office (ARO). Currently, he is the principal investigator on five active NSF grants. Dr. Xu is a recipient of NSF CAREER award, David J. Sumanth Early Career Research Award at UM, Distinguished Ph.D. dissertation Award at UCLA, and APCC Best Paper Award.