Carefully Blending Adversarial Training and Purification Improves Adversarial Robustness

Code for model training, evaluation, and reproduction of the experiments contained in the paper “Carefully Blending Adversarial Training and Purification Improves Adversarial Robustness” by Emanuele Ballarin, Alessio Ansuini and Luca Bortolussi (2024).

---

Instructions for reproducibility can be found in REPRO.md