Argus: Federated Non-convex Bilevel Learning over 6G Space-Air-Ground Integrated Network

TL;DR

This paper introduces Argus, an asynchronous federated bilevel learning algorithm designed for the dynamic and infrastructureless SAGIN environment in 6G networks, improving training efficiency and robustness.

Contribution

It proposes a novel asynchronous algorithm for non-convex federated bilevel learning over SAGIN, addressing challenges of time-varying networks and stragglers.

Findings

Theoretical analysis of iteration, communication, and computational complexities.

Numerical experiments demonstrate the algorithm's effectiveness.

Addresses non-convex, non-smooth bilevel problems in SAGIN environments.

Abstract

The space-air-ground integrated network (SAGIN) has recently emerged as a core element in the 6G networks. However, traditional centralized and synchronous optimization algorithms are unsuitable for SAGIN due to infrastructureless and time-varying environments. This paper aims to develop a novel Asynchronous algorithm a.k.a. Argus for tackling non-convex and non-smooth decentralized federated bilevel learning over SAGIN. The proposed algorithm allows networked agents (e.g. autonomous aerial vehicles) to tackle bilevel learning problems in time-varying networks asynchronously, thereby averting stragglers from impeding the overall training speed. We provide a theoretical analysis of the iteration complexity, communication complexity, and computational complexity of Argus. Its effectiveness is further demonstrated through numerical experiments.