The Spear library provides reference implementations of adversarial attacks using PyTorch. It was designed to benchmark adversarial defence methods.
The library contains implementations of the following adversarial attacks:
- Fast Gradient Sign Method (Goodfellow et al., 2014)
- DeepFool (Moosavi-Dezfooli et al., 2015)
- Basic Iterative Method (Kurakin et al., 2016)
- Carlini Wagner Attack with L2 norm (Carlini and Wagner, 2016)
- Projected Gradient Descent (Madry et al., 2017)
- Momentum Iterative - Fast Gradient Sign Method(Dong et al., 2017)