Torchattacks : A Pytorch Repository for Adversarial Attacks

Torchattacks is a PyTorch (Paszke et al. 2019) library that contains adversarial attacks to generate adversarial examples and to verify the robustness of deep learning models. Since Szegedy et al. (2013) found that deep learning models are vulnerable to the perturbed examples with small noises, called adversarial examples, various adversarial attacks have been continuously proposed. In this technical report, we provide a list of implemented adversarial attacks and explain the algorithms of each method. Here are some important things to check before generating adversarial examples. To make it easy to use adversarial attacks, a reverse-normalization is not included in the attack process.