English

Adversarial Defense by Suppressing High-frequency Components

Computer Vision and Pattern Recognition 2019-09-04 v3 Machine Learning Image and Video Processing

Abstract

Recent works show that deep neural networks trained on image classification dataset bias towards textures. Those models are easily fooled by applying small high-frequency perturbations to clean images. In this paper, we learn robust image classification models by removing high-frequency components. Specifically, we develop a differentiable high-frequency suppression module based on discrete Fourier transform (DFT). Combining with adversarial training, we won the 5th place in the IJCAI-2019 Alibaba Adversarial AI Challenge. Our code is available online.

Keywords

Cite

@article{arxiv.1908.06566,
  title  = {Adversarial Defense by Suppressing High-frequency Components},
  author = {Zhendong Zhang and Cheolkon Jung and Xiaolong Liang},
  journal= {arXiv preprint arXiv:1908.06566},
  year   = {2019}
}

Comments

3 pages. This paper is a technical report of the 5th place solution in the IJCAI-2019 Alibaba Adversarial AI Challenge. This paper has been accepted by the corresponding workshop

R2 v1 2026-06-23T10:50:26.126Z