convex_adversarial
Python
★ 391
updated 4y ago
A method for training neural networks that are provably robust to adversarial attacks.
No plain-English explanation yet — one is being written right now. Check back in a minute.