Graph-Assisted Communication-Efficient Ensemble Federated Learning

TL;DR

This paper introduces a graph-based ensemble federated learning framework that enhances communication efficiency by selectively transmitting pre-trained models and dynamically updating model confidence structures.

Contribution

It proposes a novel graph-assisted algorithm for ensemble federated learning that reduces communication costs and adapts model confidence over time.

Findings

Achieves sub-linear regret bound.

Demonstrates improved communication efficiency.

Effective on real datasets.

Abstract

Communication efficiency arises as a necessity in federated learning due to limited communication bandwidth. To this end, the present paper develops an algorithmic framework where an ensemble of pre-trained models is learned. At each learning round, the server selects a subset of pre-trained models to construct the ensemble model based on the structure of a graph, which characterizes the server's confidence in the models. Then only the selected models are transmitted to the clients, such that certain budget constraints are not violated. Upon receiving updates from the clients, the server refines the structure of the graph accordingly. The proposed algorithm is proved to enjoy sub-linear regret bound. Experiments on real datasets demonstrate the effectiveness of our novel approach.