Hierarchical Federated ADMM

Abstract--In this paper, we depart from the widely-used gradient descent-based hierarchical federated learning (FL) algorithms to develop a novel hierarchical FL framework based on the alternating direction method of multipliers (ADMM). Within this framework, we propose two novel FL algorithms, which both use ADMM in the top layer: one that employs ADMM in the lower layer and another that uses the conventional gradient descentbased approach. The proposed framework enhances privacy, and experiments demonstrate the superiority of the proposed algorithms compared to the conventional algorithms in terms Figure 1: Modular schematic of FL algorithms of learning convergence and accuracy. This conventional approach reveals the entire knowledge of the model by directly transmitting its Federated learning (FL) is gaining popularity for tasks parameters. Additionally, FL accelerates method of multipliers (ADMM) [18] within a single client set the learning process by allowing parallel computations across has been proposed [19], [20].