AI Paper Atlas

Research map / Point clouds / Point cloud classification

Robustness Certification for Point Cloud Models

Tobias Lorenz, Anian Ruoss, Mislav Balunović, Gagandeep Singh, Martin Vechev

ICCV 2021 · Point clouds › Point cloud classification · approach: Adversarial point cloud attacks · 18 citations (OpenAlex)

Read the paper

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 changedWhereMeasuredEffect
add improved max pool relaxation instead of DeepPoly (ablation)Taylor3D on PointNet · ModelNet40 Rot_Z +-3deg, 1024 pointscertified %: 28.1 -> 37.1Improved
add improved max pool relaxation instead of DeepPoly (ablation)Taylor3D on PointNet · ModelNet40 Rot_Z +-3deg, 128 pointscertified %: 72.2 -> 81.1Improved
add improved max pool relaxation (ablation)DeepPoly on MNIST CNN · MNIST l_inf eps=0.005certified images %: 72.8 -> 79.0Improved
add improved max pool relaxation (ablation)DeepPoly on MNIST CNN · MNIST l_inf eps=0.015certified images %: 3.7 -> 6.2Improved
add provable (IBP) training instead of natural training (ablation)PointNet max pool · ModelNet40 l_inf eps=0.01 (64 points)certified %: 0.0 -> 91.4Improved
add provable (IBP) training instead of natural training (ablation)PointNet max pool · ModelNet40 l_inf eps=0.01 (64 points)accuracy %: 85.7 -> 77.6Hurt
add adversarial (FGSM) training instead of natural training (ablation)PointNet max pool · ModelNet40 l_inf eps=0.01 (64 points)certified %: 0.0 -> 22.2Improved
replace max pool with average pool (ablation)PointNet natural training · ModelNet40 (64 points)accuracy %: 85.7 -> 81.9Hurt
replace max pool with average pool (ablation)PointNet adversarial training · ModelNet40 l_inf eps=0.01 (64 points)certified %: 22.2 -> 47.7Improved
increase max pool group size from 4 to 12 (ablation)DeepPoly improved max pool · ModelNet40 Rot_Z +-3degcertified %: 93.5 -> 93.5No clear effect
use Taylor3D instead of DeepG3D3DCertify · Twist o Taper o Rot_Z (10, 0.1, 1deg)running time ms: 65447ms -> 65ms (1000x speed-up)Improved
use DeepG3D instead of Taylor3D3DCertify · Twist o Rot_Z (20, 1deg)certified %: 7.6 -> 16.3Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy