Timezone: »

Beyond Pretrained Features: Noisy Image Modeling Provides Adversarial Defense
Zunzhi You · Daochang Liu · Bohyung Han · Chang Xu

Thu Dec 14 08:45 AM -- 10:45 AM (PST) @ Great Hall & Hall B1+B2 #820

Recent advancements in masked image modeling (MIM) have made it a prevailing framework for self-supervised visual representation learning. The MIM pretrained models, like most deep neural network methods, remain vulnerable to adversarial attacks, limiting their practical application, and this issue has received little research attention. In this paper, we investigate how this powerful self-supervised learning paradigm can provide adversarial robustness to downstream classifiers. During the exploration, we find that noisy image modeling (NIM), a simple variant of MIM that adopts denoising as the pre-text task, reconstructs noisy images surprisingly well despite severe corruption. Motivated by this observation, we propose an adversarial defense method, referred to as De^3, by exploiting the pretrained decoder for denoising. Through De^3, NIM is able to enhance adversarial robustness beyond providing pretrained features. Furthermore, we incorporate a simple modification, sampling the noise scale hyperparameter from random distributions, and enable the defense to achieve a better and tunable trade-off between accuracy and robustness. Experimental results demonstrate that, in terms of adversarial robustness, NIM is superior to MIM thanks to its effective denoising capability. Moreover, the defense provided by NIM achieves performance on par with adversarial training while offering the extra tunability advantage. Source code and models are available at https://github.com/youzunzhi/NIM-AdvDef.

Author Information

Zunzhi You (University of Sydney)
Zunzhi You

Zunzhi You is a first-year PhD student at USYD, supervised by Dr Chang Xu. His research interests center around the robustness and interpretability of machine learning models in computer vision applications.

Daochang Liu (The University of Sydney)
Bohyung Han (Seoul National University)
Chang Xu (University of Sydney)

More from the Same Authors