Robustness Guarantees for Adversarially Trained Neural Networks

Oct-8-2025, 23:24:36 GMT–Neural Information Processing Systems 

This allows us to give a convergence guarantee for the inner-loop PGD attack.