Boosting Adversarial Training with Hypersphere Embedding

Environment settings and libraries we used in our experiments

This project is tested under the following environment settings:

• OS: Ubuntu 16.04.3
• GPU: Geforce 2080 Ti or Tesla P100
• Cuda: 9.0, Cudnn: v7.03
• Python: 3.5.2
• PyTorch: 1.2.0
• Torchvision: 0.4.0

The guidelines

We provide the training and evaluations codes on CIFAR-10 / CIFAR-10-C in here, and those on ImageNet / ImageNet-C in here, respectively. These codes can re-implements the results in our paper.