Scholarlink

Abstract <p>Baikal-GVD is a neutrino telescope with an effective volume of approximately 0.7 km located in Lake Baikal. To facilitate observations within the framework of neutrino astronomy, two critical event processing challenges should be addressed: (1) the isolation of the neutrino-induced events from background events caused by extensive air showers, (2) the reconstruction of muon neutrino-induced event parameters, such as the energy of the corresponding muon. This report presents a neural network-based approach, to address these challenges. For the first problem, we show that convolutional neural networks can preserve 64% of neutrino-induced events while achieving the background suppression factor of 10<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11496_2025_9185_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">\(^{6}\)</EquationSource> <!--PhysPart2570080Matseiko-m1--> </InlineEquation>. This model could be instrumental in creating catalogs of neutrino-induced events. For the second problem, we develop a neural network model that reconstructs both the energy and its associated uncertainty for a given event. On Monte Carlo simulated data we achieve an error factor of 3 for a wide range of energies (from 10 GeV and above) and a factor of 2 for high energies (from 10 TeV and above). The networks were trained and evaluated using Monte Carlo simulated data.</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

Neural Networks in the Baikal-GVD Experiment: Selection of Neutrino Events and Neutrino Energy Reconstruction

  • 期刊论文

  • 发表日期：

    2025年10月25日

查看全文
  • A. V. Matseiko,
  • I. V. Kharuk

摘要

Abstract

Baikal-GVD is a neutrino telescope with an effective volume of approximately 0.7 km located in Lake Baikal. To facilitate observations within the framework of neutrino astronomy, two critical event processing challenges should be addressed: (1) the isolation of the neutrino-induced events from background events caused by extensive air showers, (2) the reconstruction of muon neutrino-induced event parameters, such as the energy of the corresponding muon. This report presents a neural network-based approach, to address these challenges. For the first problem, we show that convolutional neural networks can preserve 64% of neutrino-induced events while achieving the background suppression factor of 10 \(^{6}\) . This model could be instrumental in creating catalogs of neutrino-induced events. For the second problem, we develop a neural network model that reconstructs both the energy and its associated uncertainty for a given event. On Monte Carlo simulated data we achieve an error factor of 3 for a wide range of energies (from 10 GeV and above) and a factor of 2 for high energies (from 10 TeV and above). The networks were trained and evaluated using Monte Carlo simulated data.