ML Security PapersPWAVEP: Purifying Imperceptible Adversarial Perturbations in 3D Point Clouds via Spectral Graph Wavelets
Haoran Li 1, Renyang Liu 2, Hongjia Liu 1, Chen Wang 3, Long Yin 1, Jian Xu 1
Published on arXiv
2602.03333
Input Manipulation Attack
OWASP ML Top 10 — ML01
Key Finding
PWAVEP achieves state-of-the-art adversarial robustness on 3D point cloud classification without requiring model modification or additional training data.
PWAVEP
Novel technique introduced
Recent progress in adversarial attacks on 3D point clouds, particularly in achieving spatial imperceptibility and high attack performance, presents significant challenges for defenders. Current defensive approaches remain cumbersome, often requiring invasive model modifications, expensive training procedures or auxiliary data access. To address these threats, in this paper, we propose a plug-and-play and non-invasive defense mechanism in the spectral domain, grounded in a theoretical and empirical analysis of the relationship between imperceptible perturbations and high-frequency spectral components. Building upon these insights, we introduce a novel purification framework, termed PWAVEP, which begins by computing a spectral graph wavelet domain saliency score and local sparsity score for each point. Guided by these values, PWAVEP adopts a hierarchical strategy, it eliminates the most salient points, which are identified as hardly recoverable adversarial outliers. Simultaneously, it applies a spectral filtering process to a broader set of moderately salient points. This process leverages a graph wavelet transform to attenuate high-frequency coefficients associated with the targeted points, thereby effectively suppressing adversarial noise. Extensive evaluations demonstrate that the proposed PWAVEP achieves superior accuracy and robustness compared to existing approaches, advancing the state-of-the-art in 3D point cloud purification. Code and datasets are available at https://github.com/a772316182/pwavep
Key Contributions
- Theoretical and empirical analysis linking imperceptible adversarial perturbations in 3D point clouds to high-frequency spectral components via graph wavelets
- Hierarchical purification framework (PWAVEP) that eliminates highly salient adversarial outlier points and applies spectral filtering to attenuate adversarial high-frequency coefficients in moderately salient points
- Plug-and-play, non-invasive design requiring no model modification, retraining, or auxiliary clean data access
🛡️ Threat Analysis
Defends against adversarial input perturbations crafted to cause misclassification of 3D point clouds at inference time — a direct countermeasure to input manipulation attacks. The purification framework attenuates high-frequency adversarial noise via spectral graph wavelets before the input reaches the classifier.