Training Adversarially Robust Sparse Networks via Bayesian
                 Connectivity Sampling

                      Ozan Ozdenizci 1 2 Robert Legenstein 1

             Abstract                  Seminal work by (Szegedy et al., 2013) showed that such
                                   adversarial examples can be created via perturbations that
    Deep neural networks have been shown to be
                                   are hardly perceptible to humans, which exposed important
    susce ...