UPSET and ANGRI : Breaking High Performance Image Classifiers

Abstract

In this paper, targeted fooling of high performance image classifiers is achieved by developing two novel attack methods. The first method generates universal perturbations for target classes and the second generates image specific perturbations. Extensive experiments are conducted on MNIST and CIFAR10 datasets to provide insights about the proposed algorithms and show their effectiveness.

References in corpus (3)

• Adversarial Transformation Networks: Learning to Generate Adversarial Examples
• Universal adversarial perturbations
• Adversarial examples for generative models

Cited by in corpus (2)

• POBA-GA: Perturbation Optimized Black-Box Adversarial Attacks via Genetic Algorithm
• Moving Target Defense for Deep Visual Sensing against Adversarial Examples