Fig. 6
From: On the role of deep learning model complexity in adversarial robustness for medical images
Decision boundary projection of adversarially trained (AT) Resnet8 and Resnet50 for Chest X-ray, Dermoscopy & OCT datasets. Unperturbed samples were projected on column 1 (no attack) and perturbed samples were projected on column 2 (PGD attack) for (a, b), respectively. The complexity of adversarial trained decision boundaries increased compared to the corresponding standard trained model decision boundary in Fig. 5