Imperceptible Adversarial Attack via Invertible Neural Networks
Open Access
- 26 June 2023
- journal article
- research article
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 37 (1) , 414-424
Abstract
Adding perturbations via utilizing auxiliary gradient information or discarding existing details of the benign images are two common approaches for generating adversarial examples. Though visual imperceptibility is the desired property of adversarial examples, conventional adversarial attacks still generate traceable adversarial perturbations. In this paper, we introduce a novel Adversarial Attack via Invertible Neural Networks (AdvINN) method to produce robust and imperceptible adversarial examples. Specifically, AdvINN fully takes advantage of the information preservation property of Invertible Neural Networks and thereby generates adversarial examples by simultaneously adding class-specific semantic information of the target class and dropping discriminant information of the original class. Extensive experiments on CIFAR-10, CIFAR-100, and ImageNet-1K demonstrate that the proposed AdvINN method can produce less imperceptible adversarial images than the state-of-the-art methods and AdvINN yields more robust adversarial examples with high confidence compared to other adversarial attacks. Code is available at https://github.com/jjhuangcs/AdvINN.Keywords
All Related Versions
This publication has 4 references indexed in Scilit:
- WINNet: Wavelet-Inspired Invertible Network for Image DenoisingIEEE Transactions on Image Processing, 2022
- Invertible Image DecolorizationIEEE Transactions on Image Processing, 2021
- Threat of Adversarial Attacks on Deep Learning in Computer Vision: A SurveyIEEE Access, 2018
- Image quality assessment: from error visibility to structural similarityIEEE Transactions on Image Processing, 2004