Research map / Point clouds / Point cloud classification
LG-GAN: Label Guided Adversarial Network for Flexible Targeted Attack of Point Cloud Based Deep Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the GAN loss and discriminator (ablation) | LG · ModelNet40, PointNet with SRS defense | attack success rate: 85.0 -> 88.8 | Helped |
| Add the GAN loss and discriminator (ablation) | LG · ModelNet40, PointNet with DUP-Net defense | attack success rate: 72.0 -> 84.8 | Helped |
| Use multi-layered instead of single-layered label concatenation (ablation) | LG-GAN · ModelNet40, PointNet with DUP-Net defense | attack success rate: 37.8 -> 84.8 | Helped |
| Replace ℓ₂ reconstruction loss with Chamfer loss (ablation) | LG-GAN · ModelNet40, PointNet white-box | attack success rate: 98.3 -> 96.1 | Hurt |
| Replace ℓ₂ reconstruction loss with Chamfer loss (ablation) | LG-GAN · ModelNet40, PointNet white-box | Chamfer distance: 0.038 -> 0.137 | Hurt |
| Increase classification-loss weight α (ablation) | LG-GAN · ModelNet40, PointNet | attack success rate: 90% or so | Helped |
| Increase classification-loss weight α (ablation) | LG-GAN · ModelNet40, PointNet | visual quality | Hurt |
| Center-normalize generated point clouds (ablation) | LG-GAN · ModelNet40, PointNet white-box | attack success rate: 98.2%→82.6% | Hurt |
| Generate attacks with LG-GAN instead of IFGM | PointNet-generated attacks evaluated on PointNet++ · ModelNet40, PointNet++ black-box | attack success rate: 3.0 -> 11.6 | Helped |
| Generate attacks with LG-GAN instead of IFGM | PointNet-generated attacks evaluated on DGCNN · ModelNet40, DGCNN black-box | attack success rate: 2.6 -> 14.5 | Helped |
| Generate attacks with LG-GAN instead of IFGM | PointNet white-box attack · ModelNet40, PointNet | generation time: 0.275 -> 0.040 | Helped |
Closest papers by meaning
- CPCGAN: A Controllable 3D Point Cloud Generative Adversarial Network with Semantic Label Generating (AAAI 2021 · similarity 0.86)
- Curvature-Invariant Adversarial Attacks for 3D Point Clouds (AAAI 2024 · similarity 0.83)
- Seeing is Not Believing: Adversarial Natural Object Optimization for Hard-Label 3D Scene Attacks (CVPR 2025 · similarity 0.83)
- Target-Guided Adversarial Point Cloud Transformer Towards Recognition Against Real-world Corruptions (NeurIPS 2024 · similarity 0.83)
- Improving Robustness of 3D Point Cloud Recognition from a Fourier Perspective (NeurIPS 2024 · similarity 0.83)
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 2024 · similarity 0.83)
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.