Fine-grained Synthesis of Unrestricted Adversarial Examples

20 Nov 2019  ·  Omid Poursaeed, Tianxing Jiang, Yordanos Goshu, Harry Yang, Serge Belongie, Ser-Nam Lim ·

We propose a novel approach for generating unrestricted adversarial examples by manipulating fine-grained aspects of image generation. Unlike existing unrestricted attacks that typically hand-craft geometric transformations, we learn stylistic and stochastic modifications leveraging state-of-the-art generative models. This allows us to manipulate an image in a controlled, fine-grained manner without being bounded by a norm threshold. Our approach can be used for targeted and non-targeted unrestricted attacks on classification, semantic segmentation and object detection models. Our attacks can bypass certified defenses, yet our adversarial images look indistinguishable from natural images as verified by human evaluation. Moreover, we demonstrate that adversarial training with our examples improves performance of the model on clean images without requiring any modifications to the architecture. We perform experiments on LSUN, CelebA-HQ and COCO-Stuff as high resolution datasets to validate efficacy of our proposed approach.

PDF Abstract

Results from the Paper

  Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers.


No methods listed for this paper. Add relevant methods here