On the existence of consistent adversarial attacks in high-dimensional linear classification

Open in new window