ICLR 2022计算几何与拓扑挑战赛：设计与结果
计算几何
2022-06-28 v2
摘要
本文介绍了在ICLR 2022研讨会“几何与拓扑表示学习”中举办的微分几何与拓扑计算挑战赛。该竞赛要求参赛者提供流形上机器学习算法的实现，并需遵循开源软件Geomstats（流形部分）和Scikit-Learn（机器学习部分）或PyTorch的API。在为期两个月的赛程中，挑战赛吸引了七支队伍。本文描述了挑战赛的设计并总结了其主要发现。
引用
@article{arxiv.2206.09048,
title = {ICLR 2022 Challenge for Computational Geometry and Topology: Design and Results},
author = {Adele Myers and Saiteja Utpala and Shubham Talbar and Sophia Sanborn and Christian Shewmake and Claire Donnat and Johan Mathe and Umberto Lupo and Rishi Sonthalia and Xinyue Cui and Tom Szwagier and Arthur Pignet and Andri Bergsson and Soren Hauberg and Dmitriy Nielsen and Stefan Sommer and David Klindt and Erik Hermansen and Melvin Vaupel and Benjamin Dunn and Jeffrey Xiong and Noga Aharony and Itsik Pe'er and Felix Ambellan and Martin Hanik and Esfandiar Nava-Yazdani and Christoph von Tycowicz and Nina Miolane},
journal= {arXiv preprint arXiv:2206.09048},
year = {2022}
}