Generating Transferable 3D Adversarial Point Cloud via Random Perturbation Factorization
Keywords:CV: 3D Computer Vision, CV: Bias, Fairness & Privacy, CV: Adversarial Attacks & Robustness
AbstractRecent studies have demonstrated that existing deep neural networks (DNNs) on 3D point clouds are vulnerable to adversarial examples, especially under the white-box settings where the adversaries have access to model parameters. However, adversarial 3D point clouds generated by existing white-box methods have limited transferability across different DNN architectures. They have only minor threats in real-world scenarios under the black-box settings where the adversaries can only query the deployed victim model. In this paper, we revisit the transferability of adversarial 3D point clouds. We observe that an adversarial perturbation can be randomly factorized into two sub-perturbations, which are also likely to be adversarial perturbations. It motivates us to consider the effects of the perturbation and its sub-perturbations simultaneously to increase the transferability for sub-perturbations also contain helpful information. In this paper, we propose a simple yet effective attack method to generate more transferable adversarial 3D point clouds. Specifically, rather than simply optimizing the loss of perturbation alone, we combine it with its random factorization. We conduct experiments on benchmark dataset, verifying our method's effectiveness in increasing transferability while preserving high efficiency.
How to Cite
He, B., Liu, J., Li, Y., Liang, S., Li, J., Jia, X., & Cao, X. (2023). Generating Transferable 3D Adversarial Point Cloud via Random Perturbation Factorization. Proceedings of the AAAI Conference on Artificial Intelligence, 37(1), 764-772. https://doi.org/10.1609/aaai.v37i1.25154
AAAI Technical Track on Computer Vision I