.png){kind=icon}
.png){kind=icon}
Return
Adversarial contribution-based perturbation for transferable attack on point cloud
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2026.113438.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Existing 3D transfer attacks fail under strict perturbation constraints. • A novel adversarial contribution is formulated via graph spectral perspective. • ACA employs adversarial contribution to guide ineffective perturbation dropout. • Dynamic topological updates are integrated to maintain manifold stability. • ACA outperforms SOTA transferability while preserving visual quality.
Keywords:
Adversarial contribution
Perturbation dropout
Transferable attack
Point cloud
Manifold stability
Journal
IF:
7.6
Papers:
1.3W
Citations:
4.5W