Towards Certifying `∞ Robustness
               using Neural Networks with `∞ -dist Neurons

              Bohang Zhang 1 Tianle Cai 2 3 Zhou Lu 4 Di He 5 Liwei Wang 1 6

             Abstract                 1. Introduction
                                  Modern neural networks are usually sensitive to small, ad-
   It is well-known that standard neural networks,        versarially chosen perturbations to the inputs (Szegedy et al.,
   even with a high classification accuracy, are vul-        ...