Exploring Non-additive Randomness on ViT against Query-Based Black-Box Attacks
Jindong Gu (University of Oxford),* Fangyun Wei (Microsoft Research Asia), Philip Torr (University of Oxford), Han Hu (Microsoft Research Asia)
The 34th British Machine Vision Conference
Abstract
Deep Neural Networks can be easily fooled by small and imperceptible perturbations. The query-based black-box attack (QBBA) is able to create the perturbations using model output probabilities of image queries requiring no access to the underlying models. QBBA poses realistic threats to real-world applications. Recently, various types of robustness have been explored to defend against QBBA. In this work, we first taxonomize the stochastic defense strategies against QBBA. Following our taxonomy, we propose to explore non-additive randomness in models to defend against QBBA. Specifically, we focus on underexplored Vision Transformers based on their flexible architectures. Extensive experiments show that the proposed defense approach achieves effective defense, without much sacrifice in performance.Video
Citation
@inproceedings{Gu_2023_BMVC,
author = {Jindong Gu and Fangyun Wei and Philip Torr and Han Hu},
title = {Exploring Non-additive Randomness on ViT against Query-Based Black-Box Attacks},
booktitle = {34th British Machine Vision Conference 2023, {BMVC} 2023, Aberdeen, UK, November 20-24, 2023},
publisher = {BMVA},
year = {2023},
url = {https://papers.bmvc2023.org/0406.pdf}
}
Copyright © 2023
The British Machine Vision Conference is organised by