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

Although many efforts have been made into attack and defense on the 2D imagedomain in recent years, few methods explore the vulnerability of 3D models.Existing 3D attackers generally perform point-wise perturbation over pointclouds, resulting in deformed structures or outliers, which is easilyperceivable by humans. Moreover, their adversarial examples are generated underthe white-box setting, which frequently suffers from low success rates whentransferred to attack remote black-box models. In this paper, we study 3D pointcloud attacks from two new and challenging perspectives by proposing a novelImperceptible Transfer Attack (ITA): 1) Imperceptibility: we constrain theperturbation direction of each point along its normal vector of theneighborhood surface, leading to generated examples with similar geometricproperties and thus enhancing the imperceptibility. 2) Transferability: wedevelop an adversarial transformation model to generate the most harmfuldistortions and enforce the adversarial examples to resist it, improving theirtransferability to unknown black-box models. Further, we propose to train morerobust black-box 3D models to defend against such ITA attacks by learning morediscriminative point cloud representations. Extensive evaluations demonstratethat our ITA attack is more imperceptible and transferable thanstate-of-the-arts and validate the superiority of our defense strategy.

Imperceptible Transfer Attack and Defense on 3D Point Cloud Classification