Search papers, labs, and topics across Lattice.
The paper introduces RAIN, a framework for robust and secure aggregation in federated learning under the shuffle model of differential privacy, addressing vulnerabilities to adversarial poisoning attacks and compromised shufflers. RAIN employs sign-space aggregation to measure update consistency and limit malicious influence, along with novel secret-shared protocols for shuffling and aggregation that preserve tight Shuffle-DP guarantees. Experiments demonstrate RAIN's robustness to poisoning attacks, real-time integrity verification, and significant improvements in communication cost and aggregation speed compared to existing methods.
RAIN reconciles privacy, robustness, and verifiability in federated learning under Shuffle-DP, achieving up to 90x lower communication cost and 10x faster aggregation while defending against poisoning attacks.
Secure aggregation is a foundational building block of privacy-preserving learning, yet achieving robustness under adversarial behavior remains challenging. Modern systems increasingly adopt the shuffle model of differential privacy (Shuffle-DP) to locally perturb client updates and globally anonymize them via shuffling for enhanced privacy protection. However, these perturbations and anonymization distort gradient geometry and remove identity linkage, leaving systems vulnerable to adversarial poisoning attacks. Moreover, the shuffler, typically a third party, can be compromised, undermining security against malicious adversaries. To address these challenges, we present Robust Aggregation in Noise (RAIN), a unified framework that reconciles privacy, robustness, and verifiability under Shuffle-DP. At its core, RAIN adopts sign-space aggregation to robustly measure update consistency and limit malicious influence under noise and anonymization. Specifically, we design two novel secret-shared protocols for shuffling and aggregation that operate directly on additive shares and preserve Shuffle-DP's tight privacy guarantee. In each round, the aggregated result is verified to ensure correct aggregation and detect any selective dropping, achieving malicious security with minimal overhead. Extensive experiments across comprehensive benchmarks show that RAIN maintains strong privacy guarantees under Shuffle-DP and remains robust to poisoning attacks with negligible degradation in accuracy and convergence. It further provides real-time integrity verification with complete tampering detection, while achieving up to 90x lower communication cost and 10x faster aggregation compared with prior work.
