LHC Olympics 2020：高能物理中异常检测的社区挑战赛
高能物理 - 唯象学
2021-12-22 v1 高能物理 - 实验
数据分析、统计与概率
摘要
一种数据驱动、模型无关的对撞机新物理搜索新范式正在兴起，旨在利用异常检测与机器学习的最新突破。为在此框架内开发和基准测试新的异常检测方法，拥有标准数据集至关重要。为此，我们创建了LHC Olympics 2020，一项配有模拟对撞机事件集的社区挑战赛。这些奥运会的参与者使用研发数据集开发其方法，随后在黑盒上测试：即含有未知异常（或不含）的数据集。本文将回顾LHC Olympics 2020挑战赛，包括竞赛概述、竞赛中所部署方法描述、从经验中汲取的教训，以及对未来数据集和未来对撞机数据分析的启示。
引用
@article{arxiv.2101.08320,
title = {The LHC Olympics 2020: A Community Challenge for Anomaly Detection in High Energy Physics},
author = {Gregor Kasieczka and Benjamin Nachman and David Shih and Oz Amram and Anders Andreassen and Kees Benkendorfer and Blaz Bortolato and Gustaaf Brooijmans and Florencia Canelli and Jack H. Collins and Biwei Dai and Felipe F. De Freitas and Barry M. Dillon and Ioan-Mihail Dinu and Zhongtian Dong and Julien Donini and Javier Duarte and D. A. Faroughy and Julia Gonski and Philip Harris and Alan Kahn and Jernej F. Kamenik and Charanjit K. Khosa and Patrick Komiske and Luc Le Pottier and Pablo Martín-Ramiro and Andrej Matevc and Eric Metodiev and Vinicius Mikuni and Inês Ochoa and Sang Eon Park and Maurizio Pierini and Dylan Rankin and Veronica Sanz and Nilai Sarda and Urous Seljak and Aleks Smolkovic and George Stein and Cristina Mantilla Suarez and Manuel Szewc and Jesse Thaler and Steven Tsan and Silviu-Marian Udrescu and Louis Vaslin and Jean-Roch Vlimant and Daniel Williams and Mikaeel Yunus},
journal= {arXiv preprint arXiv:2101.08320},
year = {2021}
}
备注
108 pages, 53 figures, 3 tables