Generative Adversarial Perturbations
This addresses the vulnerability of machine learning models to adversarial attacks, offering a faster and more generalizable method compared to existing iterative approaches.
The paper tackles the problem of creating adversarial examples to fool pre-trained models by proposing novel generative models that produce image-agnostic and image-dependent perturbations, achieving high fooling rates with small perturbation norms on datasets like ImageNet and Cityscapes.
In this paper, we propose novel generative models for creating adversarial examples, slightly perturbed images resembling natural images but maliciously crafted to fool pre-trained models. We present trainable deep neural networks for transforming images to adversarial perturbations. Our proposed models can produce image-agnostic and image-dependent perturbations for both targeted and non-targeted attacks. We also demonstrate that similar architectures can achieve impressive results in fooling classification and semantic segmentation models, obviating the need for hand-crafting attack methods for each task. Using extensive experiments on challenging high-resolution datasets such as ImageNet and Cityscapes, we show that our perturbations achieve high fooling rates with small perturbation norms. Moreover, our attacks are considerably faster than current iterative methods at inference time.