Skip to yearly menu bar Skip to main content


Spotlight

Most ReLU Networks Suffer from $\ell^2$ Adversarial Perturbations

Amit Daniely · Hadas Shacham

Orals & Spotlights: Graph/Relational/Theory

Abstract: We consider ReLU networks with random weights, in which the dimension decreases at each layer. We show that for most such networks, most examples $x$ admit an adversarial perturbation at an Euclidean distance of $O\left(\frac{\|x\|}{\sqrt{d}}\right)$, where $d$ is the input dimension. Moreover, this perturbation can be found via gradient flow, as well as gradient descent with sufficiently small steps. This result can be seen as an explanation to the abundance of adversarial examples, and to the fact that they are found via gradient descent.

Chat is not available.