Skip to content
InfoResearchPeer-reviewed

Federated Quantum Learning With Adaptive Differential Privacy via Quantum Noise

Published
Record updated
View JSON

Summary

This paper proposes a quantum federated learning framework that uses inherent quantum noise to enforce differential privacy, varying the number of measurement shots and relying on depolarizing channel noise. An adaptive strategy uses a multi-armed bandit agent on each quantum device to tune the privacy-performance trade-off. The authors report a convergence analysis showing the global model reaches a bounded error floor, with simulations that they say outperform baselines in accuracy and robustness against adversarial examples.