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

Machine learning models, especially deep neural networks (DNNs), have beenshown to be vulnerable against adversarial examples which are carefully craftedsamples with a small magnitude of the perturbation. Such adversarialperturbations are usually restricted by bounding their $\mathcal{L}_p$ normsuch that they are imperceptible, and thus many current defenses can exploitthis property to reduce their adversarial impact. In this paper, we insteadintroduce "unrestricted" perturbations that manipulate semantically meaningfulimage-based visual descriptors - color and texture - in order to generateeffective and photorealistic adversarial examples. We show that thesesemantically aware perturbations are effective against JPEG compression,feature squeezing and adversarially trained model. We also show that theproposed methods can effectively be applied to both image classification andimage captioning tasks on complex datasets such as ImageNet and MSCOCO. Inaddition, we conduct comprehensive user studies to show that our generatedsemantic adversarial examples are photorealistic to humans despite largemagnitude perturbations when compared to other attacks.

Unrestricted Adversarial Examples via Semantic Manipulation