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

Localized adversarial patches aim to induce misclassification in machinelearning models by arbitrarily modifying pixels within a restricted region ofan image. Such attacks can be realized in the physical world by attaching theadversarial patch to the object to be misclassified, and defending against suchattacks is an unsolved/open problem. In this paper, we propose a generaldefense framework called PatchGuard that can achieve high provable robustnesswhile maintaining high clean accuracy against localized adversarial patches.The cornerstone of PatchGuard involves the use of CNNs with small receptivefields to impose a bound on the number of features corrupted by an adversarialpatch. Given a bounded number of corrupted features, the problem of designingan adversarial patch defense reduces to that of designing a secure featureaggregation mechanism. Towards this end, we present our robust masking defensethat robustly detects and masks corrupted features to recover the correctprediction. Notably, we can prove the robustness of our defense against anyadversary within our threat model. Our extensive evaluation on ImageNet,ImageNette (a 10-class subset of ImageNet), and CIFAR-10 datasets demonstratesthat our defense achieves state-of-the-art performance in terms of bothprovable robust accuracy and clean accuracy.

PatchGuard: A Provably Robust Defense against Adversarial Patches via Small Receptive Fields and Masking