3D Point Cloud Feature Explanations Using Gradient-Based Methods

Research output: Contribution to conferencePaper

Abstract

Explainability is an important factor to drive user trust in the use of neural networks for tasks with material impact. However, most of the work done in this area focuses on image analysis and does not take into account 3D data. We extend the saliency methods that have been shown to work on image data to deal with 3D data. We analyse the features in point clouds and voxel spaces and show that edges and corners in 3D data are deemed as important features while planar surfaces are deemed less important. The approach is model-agnostic and can provide useful information about learnt features. Driven by the insight that 3D data is inherently sparse, we visualise the features learnt by a voxel-based classification network and show that these features are also sparse and can be pruned relatively easily, leading to more efficient neural networks. Our results show that the Voxception-ResNet model can be pruned down to 5% of its parameters with negligible loss in accuracy.

Bibliographical metadata

Original languageEnglish
Publication statusAccepted/In press - 20 Mar 2020
Event2020 World Congress on Computational Intelligence: IJCNN - https://wcci2020.org/, Glasgow, United Kingdom
Event duration: 19 Jul 202024 Jul 2020

Conference

Conference2020 World Congress on Computational Intelligence
Abbreviated titleWCCI
CountryUnited Kingdom
CityGlasgow
Period19/07/2024/07/20