Stealthy Patch-Wise Backdoor Attack in 3D Point Cloud via Curvature Awareness
This addresses security vulnerabilities in 3D point cloud DNNs by improving stealthiness and efficiency for backdoor attacks, representing an incremental advance over prior methods.
The paper tackles the problem of low imperceptibility and high computational cost in 3D point cloud backdoor attacks by proposing SPBA, a patch-wise framework that uses curvature awareness to inject triggers, achieving state-of-the-art attack effectiveness and defense resistance in experiments on ModelNet40 and ShapeNetPart.
Backdoor attacks pose a severe threat to deep neural networks (DNNs) by implanting hidden backdoors that can be activated with predefined triggers to manipulate model behaviors maliciously. Existing 3D point cloud backdoor attacks primarily rely on sample-wise global modifications, which suffer from low imperceptibility. Although optimization can improve stealthiness, optimizing sample-wise triggers significantly increases computational cost. To address these limitations, we propose the Stealthy Patch-Wise Backdoor Attack (SPBA), the first patch-wise backdoor attack framework for 3D point clouds. Specifically, SPBA decomposes point clouds into local patches and employs a curvature-based imperceptibility score to guide trigger injection into visually less sensitive patches. By optimizing a unified patch-wise trigger that perturbs spectral features of selected patches, SPBA significantly enhances optimization efficiency while maintaining high stealthiness. Extensive experiments on ModelNet40 and ShapeNetPart further demonstrate that SPBA surpasses prior state-of-the-art backdoor attacks in both attack effectiveness and resistance to defense methods. The code is available at https://github.com/HazardFY/SPBA.