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 vulnerability of deep neural networks to adversarial examples, which arecrafted maliciously by modifying the inputs with imperceptible perturbations tomisled the network produce incorrect outputs, reveals the lack of robustnessand poses security concerns. Previous works study the adversarial robustness ofimage classifiers on image level and use all the pixel information in an imageindiscriminately, lacking of exploration of regions with different semanticmeanings in the pixel space of an image. In this work, we fill this gap andexplore the pixel space of the adversarial image by proposing an algorithm tolooking for possible perturbations pixel by pixel in different regions of thesegmented image. The extensive experimental results on CIFAR-10 and ImageNetverify that searching for the modified pixel in only some pixels of an imagecan successfully launch the one-pixel adversarial attacks without requiring allthe pixels of the entire image, and there exist multiple vulnerable pointsscattered in different regions of an image. We also demonstrate that theadversarial robustness of different regions on the image varies with the amountof semantic information contained.

Analyzing Adversarial Robustness of Deep Neural Networks in Pixel Space: a Semantic Perspective