Research map / Point clouds / Point cloud classification
Geometry-Aware Generation of Adversarial Point Clouds
Yuxin Wen, Jiehong Lin, Ke Chen, C. L. Philip Chen, Kui Jia
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Hausdorff distance term to the geometry-aware regularizer (ablation) | GeoA3 · ModelNet40 PointNet, SOR drop 1% | attack success rate (%): 24.51 -> 83.47 | Improved |
| add local curvature consistency term (ablation) | GeoA3 · ModelNet40 PointNet, SOR drop 1% | attack success rate (%): 53.85 -> 83.47 | Improved |
| add Chamfer distance term (ablation) | GeoA3 · ModelNet40 PointNet, SOR drop 1% | attack success rate (%): 82.17 -> 83.47 | Improved |
| add Hausdorff distance term (ablation) | GeoA3 · ModelNet40 PointNet | geometric regularity R (lower better): 0.1588 -> 0.0968 | Improved |
| add curvature consistency term (ablation) | GeoA3 · ModelNet40 PointNet | geometric regularity R (lower better): 0.1055 -> 0.0968 | Improved |
| add Chamfer distance term (ablation) | GeoA3 · ModelNet40 PointNet | geometric regularity R (lower better): 0.1020 -> 0.0968 | Improved |
| use aggressive misclassification loss instead of margin-based loss (ablation) | GeoA3 · ModelNet40 PointNet, SOR drop 20% | attack success rate (%): 4.44 -> 11.71 | Improved |
| use aggressive misclassification loss instead of margin-based loss (ablation) | GeoA3 · ModelNet40 PointNet | geometric regularity R (lower better): 0.0819 -> 0.0968 | Hurt |
| replace margin-based loss with aggressive loss (ablation) | Xiang et al. · ModelNet40 PointNet, SOR drop 1% | attack success rate (%): 1.42 -> 3.21 | Hurt |
| replace margin-based loss with aggressive loss (ablation) | Xiang et al. · ModelNet40 PointNet | geometric regularity R (lower better): 0.1772 -> 0.1956 | Hurt |
| increase number of points from 1,024 to 4,096 (ablation) | GeoA3 · ModelNet40 PointNet, SOR drop 10% | attack success rate (%): 31.58 -> 22.31 | No clear effect |
| vary lambda1 and lambda2 around defaults (ablation) | GeoA3 · ModelNet40 PointNet | attack success rate | No clear effect |
Closest papers by meaning
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 2024 · similarity 0.92)
- Explicitly Perceiving and Preserving the Local Geometric Structures for 3D Point Cloud Attack (AAAI 2024 · similarity 0.92)
- Generating 3D Adversarial Point Clouds (CVPR 2019 · similarity 0.91)
- Transferable 3D Adversarial Shape Completion using Diffusion Models (ECCV 2024 · similarity 0.91)
- Curvature-Invariant Adversarial Attacks for 3D Point Clouds (AAAI 2024 · similarity 0.90)
- Manifold Constraints for Imperceptible Adversarial Attacks on Point Clouds (AAAI 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
