LoByITFL: Low Communication Secure and Private Federated Learning

TL;DR

LoByITFL is a novel federated learning scheme that ensures strong privacy and security against Byzantine clients without sacrificing privacy guarantees, using minimal communication and a trusted initialization process.

Contribution

It introduces the first communication-efficient, information-theoretically private and secure FL scheme that maintains privacy while defending against Byzantine adversaries.

Findings

The scheme guarantees privacy and Byzantine resilience theoretically.

Experimental results demonstrate convergence of LoByITFL.

The method requires minimal communication and a one-time trusted setup.

Abstract

Privacy of the clients' data and security against Byzantine clients are key challenges in Federated Learning (FL). Existing solutions to joint privacy and security incur sacrifices on the privacy guarantee. We introduce LoByITFL, the first communication-efficient information-theoretically private and secure FL scheme that makes no sacrifices on the privacy guarantees while ensuring security against Byzantine adversaries. The key components are a small and representative dataset available to the federator, a careful modification of the FLTrust algorithm, and the one-time use of a trusted third party during an initialization period. We provide theoretical guarantees on the privacy and Byzantine resilience, as well as experimental results showing the convergence of LoByITFL.