Yet Another Intermediate-Level Attack

Qizhang Li, Yiwen Guo, Hao Chen

The transferability of adversarial examples across deep neural network (DNN) models is the crux of a spectrum of black-box attacks. In this paper, we propose a novel method to enha…