Interpreting Hidden Semantics in the Intermediate Layers of 3D Point Cloud Classification Neural Network
Not on the map. Why?
Abstract (arXiv)
Although 3D point cloud classification neural network models have been widely used, the in-depth interpretation of the activation of the neurons and layers is still a challenge. We propose a novel approach, named Relevance Flow, to interpret the hidden semantics of 3D point cloud classification neural networks. It delivers the class Relevance to the activated neurons in the intermediate layers in a back-propagation manner, and associates the activation of neurons with the input points to visualize the hidden semantics of each layer. Specially, we reveal that the 3D point cloud classification neural network has learned the plane-level and part-level hidden semantics in the intermediate layers, and utilize the normal and IoU to evaluate the consistency of both levels' hidden semantics. Besides, by using the hidden semantics, we generate the adversarial attack samples to attack 3D point cloud classifiers. Experiments show that our proposed method reveals the hidden semantics of the 3D point cloud classification neural network on ModelNet40 and ShapeNet, which can be used for the unsupervised point cloud part segmentation without labels and attacking the 3D point cloud classifiers.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 27 of its 43 references that are in the map. First posted on arXiv 12 Mar 2023.
Builds on
27 of 43 references.
References · 27
- 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
- 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,248
- 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,401
- 2017Axiomatic Attribution for Deep Networkscited by 3,384
- 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,371
- 2013Visualizing and Understanding Convolutional Networkscited by 2,290
- 2017Learning Important Features Through Propagating Activation Differencescited by 1,458
- 2014How transferable are features in deep neural networks?cited by 1,225
- 2014Striving for Simplicity: The All Convolutional Netcited by 990
- 2018PointConv: Deep Convolutional Networks on 3D Point Cloudscited by 657
- 2014Understanding Deep Image Representations by Inverting Themcited by 403
- 2020A Survey on Neural Network Interpretabilitycited by 291
15 more
- 2020A Survey on the Explainability of Supervised Machine Learningcited by 227
- 2017Interpretable Convolutional Neural Networkscited by 162
- 2018PointCloud Saliency Mapscited by 112
- 2018Interpreting CNNs via Decision Treescited by 89
- 2021Repurposing GANs for One-shot Semantic Part Segmentationcited by 88
- 2019AdvPC: Transferable Adversarial Perturbations on 3D Point Cloudscited by 84
- 2022Shape-invariant 3D Adversarial Point Cloudscited by 50
- 2017Interpreting CNN Knowledge via an Explanatory Graphcited by 49
- 2021PointGuard: Provably Robust 3D Point Cloud Classificationcited by 49
- 2020Minimal Adversarial Examples for Deep Learning on 3D Point Cloudscited by 41
- 2021Imperceptible Transfer Attack and Defense on 3D Point Cloud Classificationcited by 36
- 2017Examining CNN Representations with respect to Dataset Biascited by 27
- 2016Do semantic parts emerge in Convolutional Neural Networks?cited by 23
- 2022Exploring the Devil in Graph Spectral Domain for 3D Point Cloud Attackscited by 19
- 20203D Point Cloud Feature Explanations Using Gradient-Based Methodscited by 14
Built on by
Citing papers per year: 2025: 1.
1 paper.