G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud
Deep neural networks have been shown to produce incorrect predictions when imperceptible perturbations are introduced into the clean input. This phenomenon has garnered significant attention and extensive research in 2D images. However, related work on point clouds is still in its infancy. Current m...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/1/448 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841549351365640192 |
|---|---|
| author | Geng Chen Zhiwen Zhang Yuanxi Peng Chunchao Li Teng Li |
| author_facet | Geng Chen Zhiwen Zhang Yuanxi Peng Chunchao Li Teng Li |
| author_sort | Geng Chen |
| collection | DOAJ |
| description | Deep neural networks have been shown to produce incorrect predictions when imperceptible perturbations are introduced into the clean input. This phenomenon has garnered significant attention and extensive research in 2D images. However, related work on point clouds is still in its infancy. Current methods suffer from issues such as generated point outliers and poor attack generalization. Consequently, it is not feasible to rely solely on overall or geometry-aware attacks to generate adversarial samples. In this paper, we integrate adversarial transfer networks with the geometry-aware method to introduce adversarial loss into the attack target. A state-of-the-art autoencoder is employed, and sensitivity maps are utilized. We use the autoencoder to generate a sufficiently deceptive mask that covers the original input, adjusting the critical subset through a geometry-aware trick to distort the point cloud gradient. Our proposed approach is quantitatively evaluated in terms of the attack success rate (ASR), imperceptibility, and transferability. Compared to other baselines on ModelNet40, our method demonstrates an approximately 38% improvement in ASR for black-box transferability query attacks, with an average query count of around 7.84. Comprehensive experimental results confirm the superiority of our method. |
| format | Article |
| id | doaj-art-6663bf05b12f4045a454cb38116960b3 |
| institution | Kabale University |
| issn | 2076-3417 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj-art-6663bf05b12f4045a454cb38116960b32025-01-10T13:15:35ZengMDPI AGApplied Sciences2076-34172025-01-0115144810.3390/app15010448G&G Attack: General and Geometry-Aware Adversarial Attack on the Point CloudGeng Chen0Zhiwen Zhang1Yuanxi Peng2Chunchao Li3Teng Li4College of Computer, National University of Defense Technology, Changsha 410073, ChinaCollege of Computer, National University of Defense Technology, Changsha 410073, ChinaCollege of Computer, National University of Defense Technology, Changsha 410073, ChinaCollege of Computer, National University of Defense Technology, Changsha 410073, ChinaBeijing Institute for Advanced Study, National University of Defense Technology, Beijing 100020, ChinaDeep neural networks have been shown to produce incorrect predictions when imperceptible perturbations are introduced into the clean input. This phenomenon has garnered significant attention and extensive research in 2D images. However, related work on point clouds is still in its infancy. Current methods suffer from issues such as generated point outliers and poor attack generalization. Consequently, it is not feasible to rely solely on overall or geometry-aware attacks to generate adversarial samples. In this paper, we integrate adversarial transfer networks with the geometry-aware method to introduce adversarial loss into the attack target. A state-of-the-art autoencoder is employed, and sensitivity maps are utilized. We use the autoencoder to generate a sufficiently deceptive mask that covers the original input, adjusting the critical subset through a geometry-aware trick to distort the point cloud gradient. Our proposed approach is quantitatively evaluated in terms of the attack success rate (ASR), imperceptibility, and transferability. Compared to other baselines on ModelNet40, our method demonstrates an approximately 38% improvement in ASR for black-box transferability query attacks, with an average query count of around 7.84. Comprehensive experimental results confirm the superiority of our method.https://www.mdpi.com/2076-3417/15/1/4483Dpoint cloud classificationadversaryinvisibilitytransferability |
| spellingShingle | Geng Chen Zhiwen Zhang Yuanxi Peng Chunchao Li Teng Li G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud Applied Sciences 3D point cloud classification adversary invisibility transferability |
| title | G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud |
| title_full | G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud |
| title_fullStr | G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud |
| title_full_unstemmed | G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud |
| title_short | G&G Attack: General and Geometry-Aware Adversarial Attack on the Point Cloud |
| title_sort | g g attack general and geometry aware adversarial attack on the point cloud |
| topic | 3D point cloud classification adversary invisibility transferability |
| url | https://www.mdpi.com/2076-3417/15/1/448 |
| work_keys_str_mv | AT gengchen ggattackgeneralandgeometryawareadversarialattackonthepointcloud AT zhiwenzhang ggattackgeneralandgeometryawareadversarialattackonthepointcloud AT yuanxipeng ggattackgeneralandgeometryawareadversarialattackonthepointcloud AT chunchaoli ggattackgeneralandgeometryawareadversarialattackonthepointcloud AT tengli ggattackgeneralandgeometryawareadversarialattackonthepointcloud |