Scholarlink

Abstract <p>This work presents the results of a gradient-boosted decision trees application to the particle identification problem in the MPD experiment at the NICA accelerator complex of the Joint Institute for Nuclear Research. Particle identification is one of the significant tasks at the MPD experiment. Since particle identification data is structured in general and at MPD in particular, gradient boosting algorithms were considered. Four gradient boosting implementations were considered within this work. Numerical studies were conducted to determine the differences between them. A comparison of accuracy and speed was made on Monte Carlo data. It was generated with minimum bias, bismuth and bismuth (Bi + Bi) collisions at <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11497_2025_10110_Article_IEq1.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="92" /> </InlineMediaObject> <EquationSource Format="TEX">\(\sqrt {{{s}_{{NN}}}} = 9.2\)</EquationSource> <!--PhysPNLt2570025Papoyan-m1--> </InlineEquation> GeV, which is expected to be the first colliding system at MPD. The results obtained demonstrate what kind of algorithms will be the most suitable according to the computing conditions of a physical experiment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Gradient Boosted Decision Tree for Particle Identification Problem at MPD

  • 期刊论文

  • 发表日期：

    2025年6月1日

查看全文
  • V. Papoyan,
  • A. Aparin,
  • A. Ayriyan,
  • H. Grigorian,
  • A. Korobitsin

摘要

Abstract

This work presents the results of a gradient-boosted decision trees application to the particle identification problem in the MPD experiment at the NICA accelerator complex of the Joint Institute for Nuclear Research. Particle identification is one of the significant tasks at the MPD experiment. Since particle identification data is structured in general and at MPD in particular, gradient boosting algorithms were considered. Four gradient boosting implementations were considered within this work. Numerical studies were conducted to determine the differences between them. A comparison of accuracy and speed was made on Monte Carlo data. It was generated with minimum bias, bismuth and bismuth (Bi + Bi) collisions at \(\sqrt {{{s}_{{NN}}}} = 9.2\) GeV, which is expected to be the first colliding system at MPD. The results obtained demonstrate what kind of algorithms will be the most suitable according to the computing conditions of a physical experiment.