PHY-Fed: An Information-Theoretic Secure Aggregation in Federated Learning in Wireless Communications

TL;DR

PHY-Fed introduces an information-theoretic secure aggregation framework for federated learning over wireless communications, enhancing privacy without sacrificing accuracy compared to existing methods.

Contribution

It proposes PHY-Fed, a novel framework that provides information-theoretic security for federated learning in wireless environments, addressing limitations of current privacy-preserving techniques.

Findings

Provides information-theoretic security guarantees

Maintains high test accuracy with multiple parties

Addresses vulnerabilities of existing privacy methods

Abstract

Federated learning (FL) is a type of distributed machine learning at the wireless edge that preserves the privacy of clients' data from adversaries and even the central server. Existing federated learning approaches either use (i) secure multiparty computation (SMC) which is vulnerable to inference or (ii) differential privacy which may decrease the test accuracy given a large number of parties with relatively small amounts of data each. To tackle the problem with the existing methods in the literature, In this paper, we introduce PHY-Fed, a new framework that secures federated algorithms from an information-theoretic point of view.