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 of adversarial examples is of central importance totransfer-based black-box adversarial attacks. Previous works for generatingtransferable adversarial examples focus on attacking \emph{given} pretrainedsurrogate models while the connections between surrogate models and adversarialtrasferability have been overlooked. In this paper, we propose {\em LipschitzRegularized Surrogate} (LRS) for transfer-based black-box attacks, a novelapproach that transforms surrogate models towards favorable adversarialtransferability. Using such transformed surrogate models, any existingtransfer-based black-box attack can run without any change, yet achieving muchbetter performance. Specifically, we impose Lipschitz regularization on theloss landscape of surrogate models to enable a smoother and more controlledoptimization process for generating more transferable adversarial examples. Inaddition, this paper also sheds light on the connection between the innerproperties of surrogate models and adversarial transferability, where threefactors are identified: smaller local Lipschitz constant, smoother losslandscape, and stronger adversarial robustness. We evaluate our proposed LRSapproach by attacking state-of-the-art standard deep neural networks anddefense models. The results demonstrate significant improvement on the attacksuccess rates and transferability. Our code is available athttps://github.com/TrustAIoT/LRS.

LRS: Enhancing Adversarial Transferability through Lipschitz Regularized Surrogate