How Powerful Are Randomly Initialized Pointcloud Set Functions?
Computer Vision and Pattern Recognition
2020-03-12 v1 Machine Learning
Abstract
We study random embeddings produced by untrained neural set functions, and show that they are powerful representations which well capture the input features for downstream tasks such as classification, and are often linearly separable. We obtain surprising results that show that random set functions can often obtain close to or even better accuracy than fully trained models. We investigate factors that affect the representative power of such embeddings quantitatively and qualitatively.
Cite
@article{arxiv.2003.05410,
title = {How Powerful Are Randomly Initialized Pointcloud Set Functions?},
author = {Aditya Sanghi and Pradeep Kumar Jayaraman},
journal= {arXiv preprint arXiv:2003.05410},
year = {2020}
}
Comments
6 pages, 2 figures, NeurIPS 2019 Sets & Partitions Workshop