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

The transferability and robustness of adversarial examples are two practicalyet important properties for black-box adversarial attacks. In this paper, weexplore effective mechanisms to boost both of them from the perspective ofnetwork hierarchy, where a typical network can be hierarchically divided intooutput stage, intermediate stage and input stage. Since over-specialization ofsource model, we can hardly improve the transferability and robustness of theadversarial perturbations in the output stage. Therefore, we focus on theintermediate and input stages in this paper and propose a transferable androbust adversarial perturbation generation (TRAP) method. Specifically, wepropose the dynamically guided mechanism to continuously calculate accuratedirectional guidances for perturbation generation in the intermediate stage. Inthe input stage, instead of the single-form transformation augmentationsadopted in the existing methods, we leverage multiform affine transformationaugmentations to further enrich the input diversity and boost the robustnessand transferability of the adversarial perturbations. Extensive experimentsdemonstrate that our TRAP achieves impressive transferability and highrobustness against certain interferences.

Exploring Transferable and Robust Adversarial Perturbation Generation from the Perspective of Network Hierarchy