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

Neural networks are known to be vulnerable to carefully crafted adversarialexamples, and these malicious samples often transfer, i.e., they maintain theireffectiveness even against other models. With great efforts delved into thetransferability of adversarial examples, surprisingly, less attention has beenpaid to its impact on real-world deep learning deployment. In this paper, weinvestigate the transferability of adversarial examples across a wide range ofreal-world computer vision tasks, including image classification, explicitcontent detection, optical character recognition (OCR), and object detection.It represents the cybercriminal's situation where an ensemble of differentdetection mechanisms need to be evaded all at once. We propose practical attackthat overcomes existing attacks' limitation of requiring task-specific lossfunctions by targeting on the `dispersion' of internal feature map. We reportevaluation on four different computer vision tasks provided by Google CloudVision APIs to show how our approach outperforms existing attacks by degradingperformance of multiple CV tasks by a large margin with only modestperturbations.

Enhancing Cross-task Transferability of Adversarial Examples with Dispersion Reduction