Rethinking Invariance Regularization in Adversarial Training to Improve Robustness-Accuracy Trade-off

Open in new window