• home
• current students
• graduate
• events

Presenter: Lei Zhao
Supervisor:

Date: Mon, August 21, 2023
Time: 10:00:00 - 00:00:00
Place: EOW 430

ABSTRACT

Abstract:

Driven by the increasing demand for real-time, low latency learning processes and the ever-growing emphasis on data privacy, Federated Learning (FL) enabled edge intelligence emerges as a potent and decentralized learning paradigm at the edge of the network. This approach empowers collaborative model training on edge agents, allowing them to make intelligent decisions locally without relying solely on centralized cloud servers. To enhance the training efficiency of edge agents and alleviate communication burdens, we propose a novel technique called Conjugation with inexact Line Search enabled Federated Stochastic Variance Reduced Gradient (CLSFSVRG).This method involves randomly selecting a small portion of intelligent edge agents in each round to collaboratively train their models using conjugated central acceleration and in exact global line search. Our simulations demonstrate that the proposed scheme outperforms state-of-the-art FL algorithms, achieving superior performance in terms of higher test accuracy and faster convergence speed. Remarkably, this approach reduces communication costs by an impressive 84%, while still achieving a test accuracy of 96%. Allowing a small portion of edge agents to participate, the proposed method exhibits higher robustness without compromising the achieved test accuracy. Moreover, the fast convergence speed achieved with only a limited number of participated edge agents contributes to significant reductions in edge energy consumption during the training procedure.