Adversarially Robust Medical Classification via Attentive Convolutional Neural Networks

Isaac Wasserman

Convolutional neural network-based medical image classifiers have been shown to be especially susceptible to adversarial examples. Such instabilities are likely to be unacceptable…