- Keywords: 3D Point Cloud Processing, Interpretability, Deep Learning
- TL;DR: We diagnose deep neural networks for 3D point cloud processing to explore the utility of different network architectures.
- Abstract: In this paper, we diagnose deep neural networks for 3D point cloud processing to explore the utility of different network architectures. We propose a number of hypotheses on the effects of specific network architectures on the representation capacity of DNNs. In order to prove the hypotheses, we design five metrics to diagnose various types of DNNs from the following perspectives, information discarding, information concentration, rotation robustness, adversarial robustness, and neighborhood inconsistency. We conduct comparative studies based on such metrics to verify the hypotheses, which may shed new lights on the architectural design of neural networks. Experiments demonstrated the effectiveness of our method. The code will be released when this paper is accepted.