Robustifying models against adversarial attacks by Langevin dynamics

Vignesh Srinivasan, Csaba Rohrer, Arturo Marbán, Klaus-Robert Müller, Wojciech Samek, Shinichi Nakajima