Efficient Quantum Algorithm for Robust Training

Yue Wang, Guangyi He, Liepeng Zhang +2

Adversarial training is a standard defense against malicious input perturbations in security-critical machine-learning systems. Its main burden is structural: before every paramete…