Data Augmentation for Few-Shot

Adversarial Augmentation

Augmentation technique that uses adversarial perturbations to create robust samples that improve the model's resistance to attacks. This approach strengthens generalization by exposing the model to extreme but plausible variations.

← Geri