English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The adversarial patch attack against image classification models aims toinject adversarially crafted pixels within a restricted image region (i.e., apatch) for inducing model misclassification. This attack can be realized in thephysical world by printing and attaching the patch to the victim object; thus,it imposes a real-world threat to computer vision systems. To counter thisthreat, we design PatchCleanser as a certifiably robust defense againstadversarial patches. In PatchCleanser, we perform two rounds of pixel maskingon the input image to neutralize the effect of the adversarial patch. Thisimage-space operation makes PatchCleanser compatible with any state-of-the-artimage classifier for achieving high accuracy. Furthermore, we can prove thatPatchCleanser will always predict the correct class labels on certain imagesagainst any adaptive white-box attacker within our threat model, achievingcertified robustness. We extensively evaluate PatchCleanser on the ImageNet,ImageNette, CIFAR-10, CIFAR-100, SVHN, and Flowers-102 datasets and demonstratethat our defense achieves similar clean accuracy as state-of-the-artclassification models and also significantly improves certified robustness fromprior works. Remarkably, PatchCleanser achieves 83.9% top-1 clean accuracy and62.1% top-1 certified robust accuracy against a 2%-pixel square patch anywhereon the image for the 1000-class ImageNet dataset.

PatchCleanser: Certifiably Robust Defense against Adversarial Patches for Any Image Classifier