Perfect Privacy for Discriminator-Based Byzantine-Resilient Federated Learning

2025-06-16

AI Summary

Federated learning faces dual security threats: Byzantine attacks and eavesdropping. This paper proposes ByITFL and LoByITFL—two novel frameworks that jointly achieve Byzantine robustness and information-theoretically perfect privacy protection within a single unified design. ByITFL leverages representative datasets, discriminative functions, and Lagrange coding to detect and suppress malicious model updates. LoByITFL enhances scalability by introducing re-randomization and lightweight aggregation, reducing communication overhead by over 60% with only a single trusted third-party initialization. We provide rigorous theoretical guarantees for privacy (strict information-theoretic secrecy), robustness (tolerance to arbitrary Byzantine clients), and convergence (sublinear convergence rate under standard assumptions). Extensive experiments demonstrate that both schemes significantly outperform state-of-the-art baselines across diverse Byzantine attack scenarios: ByITFL achieves provably perfect privacy, while LoByITFL delivers comparable security with drastically reduced communication cost.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Safety and Robustness
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Security and Privacy: Privacy-enhancing technologies
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Federated learning (FL) shows great promise in large-scale machine learning but introduces new privacy and security challenges. We propose ByITFL and LoByITFL, two novel FL schemes that enhance resilience against Byzantine users while keeping the users' data private from eavesdroppers. To ensure privacy and Byzantine resilience, our schemes build on having a small representative dataset available to the federator and crafting a discriminator function allowing the mitigation of corrupt users' contributions. ByITFL employs Lagrange coded computing and re-randomization, making it the first Byzantine-resilient FL scheme with perfect Information-Theoretic (IT) privacy, though at the cost of a significant communication overhead. LoByITFL, on the other hand, achieves Byzantine resilience and IT privacy at a significantly reduced communication cost, but requires a Trusted Third Party, used only in a one-time initialization phase before training. We provide theoretical guarantees on privacy and Byzantine resilience, along with convergence guarantees and experimental results validating our findings.

Problem

• Enhance Byzantine resilience in federated learning
• Ensure perfect privacy for user data
• Reduce communication overhead in secure FL

Innovation

• Lagrange coded computing for perfect IT privacy
• Discriminator function to mitigate corrupt users
• Trusted Third Party for reduced communication cost