Quantum Adversarial Learning for Kernel Methods

arXiv:2404.05824v1 Announce Type: cross
Abstract: We show that hybrid quantum classifiers based on quantum kernel methods and support vector machines are vulnerable against adversarial attacks, namely small engineered perturbations of the input data can deceive the classifier into predicting the wrong result. Nonetheless, we also show that simple defence strategies based on data augmentation with a few crafted perturbations can make the classifier robust against new attacks. Our results find applications in security-critical learning problems and in mitigating the effect …

abstract adversarial adversarial attacks adversarial learning arxiv attacks augmentation classifier classifiers cs.cr cs.lg data defence hybrid kernel machines quant-ph quantum show simple small strategies support support vector machines type vector vulnerable