arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.16074cs.LGcs.CRcs.CV
收藏

PointBA:迈向三维点云中的后门攻击

PointBA: Towards Backdoor Attacks in 3D Point Cloud

National University of Singapore(新加坡国立大学);Southwest Jiaotong University(西南交通大学);Institute of High Performance Computing, A*STAR(高性能计算研究所,A*STAR)

Xinke Li, Zhirui Chen, Yue Zhao, Zekun Tong, Yabang Zhao, Andrew Lim, Joey Tianyi Zhou

AI总结:

本文首次系统研究三维点云后门攻击,提出毒标签攻击 PointPBA 和清洁标签攻击 PointCBA,利用三维数据特性与特征解耦技术实现攻击,实验成功率分别超过 95% 和约 50%,为三维深度模型鲁棒性提供基准。

摘要:

三维深度学习在包括许多安全关键型应用在内的多种任务中日益流行。然而,近期一些工作揭示了三维深度模型的安全问题。尽管其中大多数工作关注对抗攻击,但我们认为后门攻击实际上是对三维深度学习系统更严重的威胁,且目前尚未被探索。我们利用三维数据和网络的独特性质,在一个统一框架中提出了三维点云中的后门攻击。具体而言,我们设计了两种针对点云的攻击方法:毒标签后门攻击(PointPBA)和清洁标签后门攻击(PointCBA)。前者在实际中直接且有效,而后者在假设存在一定数据审查的情况下更为隐蔽。攻击算法的提出和发展主要受到以下两方面启发:1)近期关于三维对抗样本的发现表明深度模型在空间变换下具有脆弱性;2)所提出的特征解耦技术通过优化方法操纵数据特征,并具有嵌入新任务的潜力。大量实验表明,PointPBA 在各种三维数据集和模型上的成功率超过 95%,而更隐蔽的 PointCBA 成功率约为 50%。我们提出的三维点云后门攻击有望作为提升三维深度模型鲁棒性的基准。

英文摘要:

3D deep learning has been increasingly more popular for a variety of tasks including many safety-critical applications. However, recently several works raise the security issues of 3D deep models. Although most of them consider adversarial attacks, we identify that backdoor attack is indeed a more serious threat to 3D deep learning systems but remains unexplored. We present the backdoor attacks in 3D point cloud with a unified framework that exploits the unique properties of 3D data and networks. In particular, we design two attack approaches on point cloud: the poison-label backdoor attack (PointPBA) and the clean-label backdoor attack (PointCBA). The first one is straightforward and effective in practice, while the latter is more sophisticated assuming there are certain data inspections. The attack algorithms are mainly motivated and developed by 1) the recent discovery of 3D adversarial samples suggesting the vulnerability of deep models under spatial transformation; 2) the proposed feature disentanglement technique that manipulates the feature of the data through optimization methods and its potential to embed a new task. Extensive experiments show the efficacy of the PointPBA with over 95% success rate across various 3D datasets and models, and the more stealthy PointCBA with around 50% success rate. Our proposed backdoor attack in 3D point cloud is expected to perform as a baseline for improving the robustness of 3D deep models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑