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 are widely known to be susceptible to adversarialexamples, which can cause incorrect predictions through subtle inputmodifications. These adversarial examples tend to be transferable betweenmodels, but targeted attacks still have lower attack success rates due tosignificant variations in decision boundaries. To enhance the transferabilityof targeted adversarial examples, we propose introducing competition into theoptimization process. Our idea is to craft adversarial perturbations in thepresence of two new types of competitor noises: adversarial perturbationstowards different target classes and friendly perturbations towards the correctclass. With these competitors, even if an adversarial example deceives anetwork to extract specific features leading to the target class, thisdisturbance can be suppressed by other competitors. Therefore, within thiscompetition, adversarial examples should take different attack strategies byleveraging more diverse features to overwhelm their interference, leading toimproving their transferability to different models. Considering thecomputational complexity, we efficiently simulate various interference fromthese two types of competitors in feature space by randomly mixing up storedclean features in the model inference and named this method Clean Feature Mixup(CFM). Our extensive experimental results on the ImageNet-Compatible andCIFAR-10 datasets show that the proposed method outperforms the existingbaselines with a clear margin. Our code is available athttps://github.com/dreamflake/CFM.

Introducing Competition to Boost the Transferability of Targeted Adversarial Examples through Clean Feature Mixup