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

Deep neural networks (DNNs) can be manipulated to exhibit specific behaviorswhen exposed to specific trigger patterns, without affecting their performanceon benign samples, dubbed backdoor attack. Some recent research has focused ondesigning invisible triggers for backdoor attacks to ensure visualstealthiness, while showing high effectiveness, even under backdoor defense.However, we find that these carefully designed invisible triggers are oftensensitive to visual distortion during inference, such as Gaussian blurring orenvironmental variations in physical scenarios. This phenomenon couldsignificantly undermine the practical effectiveness of attacks, but has beenrarely paid attention to and thoroughly investigated. To address thislimitation, we define a novel trigger called the Visible, Semantic,Sample-Specific, and Compatible trigger (VSSC trigger), to achieve effective,stealthy and robust to visual distortion simultaneously. To implement it, wedevelop an innovative approach by utilizing the powerful capabilities of largelanguage models for choosing the suitable trigger and text-guided image editingtechniques for generating the poisoned image with the trigger. Extensiveexperimental results and analysis validate the effectiveness, stealthiness androbustness of the VSSC trigger. It demonstrates superior robustness todistortions compared with most digital backdoor attacks and allows moreefficient and flexible trigger integration compared to physical backdoorattacks. We hope that the proposed VSSC trigger and implementation approachcould inspire future studies on designing more practical triggers in backdoorattacks.

Robust Backdoor Attack with Visible, Semantic, Sample-Specific, and Compatible Triggers