Model Extraction Attacks in GAN-Based Image Translation: Mitigating Domain Shift

The author introduces a novel approach to model extraction attacks in image translation tasks by addressing the domain shift problem through wavelet regularization and sharpness-aware minimization.