Random Walks for Adversarial Meshes
Not on the map. Why?
Abstract (arXiv)
A polygonal mesh is the most-commonly used representation of surfaces in computer graphics. Therefore, it is not surprising that a number of mesh classification networks have recently been proposed. However, while adversarial attacks are wildly researched in 2D, the field of adversarial meshes is under explored. This paper proposes a novel, unified, and general adversarial attack, which leads to misclassification of several state-of-the-art mesh classification neural networks. Our attack approach is black-box, i.e. it has access only to the network's predictions, but not to the network's full architecture or gradients. The key idea is to train a network to imitate a given classification network. This is done by utilizing random walks along the mesh surface, which gather geometric information. These walks provide insight onto the regions of the mesh that are important for the correct prediction of the given classification network. These mesh regions are then modified more than other regions in order to attack the network in a manner that is barely visible to the naked eye.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 34 of its 45 references that are in the map. First posted on arXiv 15 Feb 2022.
Builds on
34 of 45 references.
References · 34
- 2014Explaining and Harnessing Adversarial Examplescited by 7,364
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
- 2013Intriguing properties of neural networkscited by 5,155
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
- 2016Convolutional Neural Networks on Graphs with Fast Localized Spectral Filteringcited by 2,089
- 2016Practical Black-Box Attacks against Machine Learningcited by 903
- 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 722
- 2017Evasion Attacks against Machine Learning at Test Timecited by 689
- 2018Improving Transferability of Adversarial Examples with Input Diversitycited by 599
- 20163D Menagerie: Modeling the 3D shape and pose of animalscited by 312
- 2018Audio Adversarial Examples: Targeted Attacks on Speech-to-Textcited by 274
- 2020Understanding and Improving Fast Adversarial Trainingcited by 214
22 more
- 2018Generating 3D Adversarial Point Cloudscited by 144
- 2018Decoupling Direction and Norm for Efficient Gradient-Based L2 Adversarial Attacks and Defensescited by 109
- 2019PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Cloudscited by 99
- 2018MeshNet: Mesh Neural Network for 3D Shape Representationcited by 95
- 2019Sparse and Imperceivable Adversarial Attackscited by 95
- 2019Extending Adversarial Attacks and Defenses to Deep 3D Point Cloud Classifierscited by 92
- 2019AdvPC: Transferable Adversarial Perturbations on 3D Point Cloudscited by 84
- 2018MeshCNN: A Network with an Edgecited by 69
- 2019Efficient Adversarial Training with Transferable Adversarial Examplescited by 56
- 2019Robustness of 3D Deep Learning in an Adversarial Settingcited by 55
- 2020On Isometry Robustness of Deep 3D Point Cloud Models under Adversarial Attackscited by 46
- 2020IF-Defense: 3D Adversarial Point Cloud Defense via Implicit Function based Restorationcited by 46
- 2020Primal-Dual Mesh Convolutional Neural Networkscited by 44
- 2019Data Poisoning Attack against Knowledge Graph Embeddingcited by 36
- 2020MeshWalker: Deep Mesh Understanding by Random Walkscited by 32
- 20213D Adversarial Attacks Beyond Point Cloudcited by 26
- 2018Data Poisoning Attack against Unsupervised Node Embedding Methodscited by 19
- 2021Universal Spectral Adversarial Attacks for Deformable Shapescited by 14
- 2018Gray-box Adversarial Trainingcited by 10
- 2017Practical Attacks Against Graph-based Clusteringcited by 9
- 2021AttWalk: Attentive Cross-Walks for Deep Mesh Analysiscited by 5
- 2018MeshAdv: Adversarial Meshes for Visual Recognitioncited by 3
Built on by
Citing papers per year: 2022: 1; 2023: 0; 2024: 1; 2025: 0; 2026: 1.
3 papers.