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

Development of defenses against physical world attacks such as adversarialpatches is gaining traction within the research community. We contribute to thefield of adversarial patch detection by introducing an uncertainty-basedadversarial patch localizer which localizes adversarial patch on an image,permitting post-processing patch-avoidance or patch-reconstruction. We quantifyour prediction uncertainties with the development of \textit{\textbf{D}etectionof \textbf{U}ncertainties in the \textbf{E}xceedance of \textbf{T}hreshold}(DUET) algorithm. This algorithm provides a framework to ascertain confidencein the adversarial patch localization, which is essential for safety-sensitiveapplications such as self-driving cars and medical imaging. We conductedexperiments on localizing adversarial patches and found our proposed DUET modeloutperforms baseline models. We then conduct further analyses on our choice ofmodel priors and the adoption of Bayesian Neural Networks in different layerswithin our model architecture. We found that isometric gaussian priors inBayesian Neural Networks are suitable for patch localization tasks and thepresence of Bayesian layers in the earlier neural network blocks facilitatestop-end localization performance, while Bayesian layers added in the laterneural network blocks contribute to better model generalization. We thenpropose two different well-performing models to tackle different use cases.

Detection of Uncertainty in Exceedance of Threshold (DUET): An Adversarial Patch Localizer