Engineering Commons^™

Gradient Free Sign Activation Zero One Loss Neural Networks For Adversarially Robust Classification, Yunzhe Xue

Dissertations

The zero-one loss function is less sensitive to outliers than convex surrogate losses such as hinge and cross-entropy. However, as a non-convex function, it has a large number of local minima, andits undifferentiable attribute makes it impossible to use backpropagation, a method widely used in training current state-of-the-art neural networks. When zero-one loss is applied to deep neural networks, the entire training process becomes challenging. On the other hand, a massive non-unique solution probably also brings different decision boundaries when optimizing zero-one loss, making it possible to fight against transferable adversarial examples, which is a common weakness in deep learning …