Efficient Secure Aggregation for Privacy-Preserving Federated Machine Learning

TL;DR

This paper introduces e-SeaFL, a highly efficient secure aggregation protocol for federated learning that reduces communication rounds and computational costs, enabling privacy-preserving model updates with large gradient vectors.

Contribution

e-SeaFL is the first protocol to achieve one-round secure aggregation with verifiable proofs using homomorphic vector commitments, improving efficiency for large models.

Findings

Order of magnitude efficiency improvement over previous protocols.

Supports large gradient vectors with thousands of parameters.

Open-source implementation available for practical use.

Abstract

Secure aggregation protocols ensure the privacy of users' data in federated learning by preventing the disclosure of local gradients. Many existing protocols impose significant communication and computational burdens on participants and may not efficiently handle the large update vectors typical of machine learning models. Correspondingly, we present e-SeaFL, an efficient verifiable secure aggregation protocol taking only one communication round during the aggregation phase. e-SeaFL allows the aggregation server to generate proof of honest aggregation to participants via authenticated homomorphic vector commitments. Our core idea is the use of assisting nodes to help the aggregation server, under similar trust assumptions existing works place upon the participating users. Our experiments show that the user enjoys an order of magnitude efficiency improvement over the state-of-the-art (IEEE S\&P 2023) for large gradient vectors with thousands of parameters. Our open-source implementation is available at https://github.com/vt-asaplab/e-SeaFL.