Byzantine-Robust Federated Learning

Overview

Byzantine-robust federated learning studies how to train a shared model when some participating clients may be faulty or malicious, without centralizing their data. Common defenses replace simple averaging with coordinate medians or trimmed means, Krum-style distance selection, geometric-median or clustering methods, and trust, anomaly, or history-based screening; newer systems also combine these with secure aggregation, differential privacy, or proofs of correct computation. Methods are tested on non-IID image and text benchmarks, intrusion detection and medical data, and increasingly decentralized, asynchronous, or edge settings, where partial participation and client diversity complicate detection. The central tension is separating malicious updates from legitimate variation while retaining accuracy and convergence: adaptive or coordinated attacks can evade filters, and guarantees often depend on assumptions about attacker numbers, honest-client structure, or available reference data.

Latest papers

91 papers7 papers in the last four weeks, down 59% on the four weeks before. 0.1% of all new papers.

Oct 6, 2023

Kick Bad Guys Out! Conditionally Activated Anomaly Detection in Federated Learning with Zero-Knowledge Proof Verification

Federated Learning (FL) systems are susceptible to adversarial attacks, such as model poisoning attacks and backdoor attacks. Existing defense mechanisms face critical limitations in deployments, such as relying on impractical assumptions (e.g., adversaries acknowledging the presence of attacks before attacking) or undermining accuracy in model training, even in benign scenarios. To address these challenges, we propose CustodianFL, a two-staged anomaly detection method specifically designed for FL deployments. In the first stage, it flags suspicious client activities. In the second stage that is activated only when needed, it further examines these candidates using Three-Sigma Rule to identify and exclude truly malicious local models from FL training. To ensure integrity and transparency within the FL system, CustodianFL integrates zero-knowledge proofs, enabling clients to cryptographically verify the server's detection process without relying on the server's goodwill. CustodianFL operates without unrealistic assumptions and avoids interfering with FL training in attack-free scenarios. It bridges the gap between theoretical advances in FL security and the practical demands of real FL systems. Experimental results demonstrate that CustodianFL consistently delivers performance comparable to benign cases, highlighting its effectiveness in identifying and eliminating malicious models with high accuracy.