Quantum Secure Private Inference from Vacuum Fluctuations

Kfir Sulimany, Sri Krishna Vadlamani, Ryan Hamerly, Prahlad Iyengar, Dirk Englund The authors use the vacuum fluctuations of coherent light as a cryptographic resource for private neural-network inference. They develop security bounds for model and data leakage and demonstrate experimentally that the approach can achieve high MNIST accuracy using hardware similar to that used in CV-QKD