Improving the machine learning based vertex reconstruction for large liquid scintillator detectors with multiple types of PMTs

Li, Zi-Yuan; Zhang, Qian; He, Jie-Han; Wang, He; Wu, Chengxin; Cai, X.; Zhou, You; Zhang, Yumei; Luo, W.

doi:10.48550/arxiv.2205.04039

Cited by 2 publications

(2 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the subsequent paper we, using an optimal subset from a large set of newly engineered aggregated features, demonstrated that this approach can achieve the same performance as the PMT-wise one [13]. On the other hand, vertex reconstruction requires usage of granular information both with traditional and ML algorithms, see [14,15]. Actual research aims to further investigate the potential of the aggregated features approach and study Boosted Decision Trees and Fully Connected Deep Neural Network techniques for energy reconstruction in JUNO.…”

Section: Problem Statementmentioning

confidence: 87%

Energy reconstruction for large liquid scintillator detectors with machine learning techniques: aggregated features approach

Gavrikov,

Malyshkin,

Ratnikov

2022

Preprint

View full text Add to dashboard Cite

Large scale detectors consisting of a liquid scintillator (LS) target surrounded by an array of photo-multiplier tubes (PMT) are widely used in modern neutrino experiments: Borexino, KamLAND, Daya Bay, Double Chooz, RENO, and upcoming JUNO with its satellite detector TAO. Such apparatuses are able to measure neutrino energy, which can be derived from the amount of light and its spatial and temporal distribution over PMT-channels. However, achieving a fine energy resolution in large scale detectors is challenging.In this work, we present machine learning methods for energy reconstruction in JUNO, the most advanced detector of its type. We focus on positron events in the energy range of 0-10 MeV which corresponds to the main signal in JUNO -neutrinos originated from nuclear reactor cores and detected via an inverse beta-decay channel. We consider Boosted Decision Trees and Fully Connected Deep Neural Network trained on aggregated features, calculated using information collected by PMTs. We describe the details of our feature engineering procedure and show that machine learning models can provide energy resolution 𝜎 = 3% at 1 MeV using subsets of engineered features. The dataset for model training and testing is generated by the Monte Carlo method with the official JUNO software. Consideration of calibration sources for evaluation of the reconstruction algorithms performance on real data is also presented.

show abstract

Section: Problem Statementmentioning

confidence: 87%

Energy reconstruction for large liquid scintillator detectors with machine learning techniques: aggregated features approach

Gavrikov,

Malyshkin,

Ratnikov

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…It describes detector information through a set of tags and attributes in plain text format to provide persistent detector description for an experiment. Several HEP experiments, including BESIII [8,9], PHENIX [21], LHCb [22][23][24] and JUNO [25], have used GDML to describe and optimize the detector geometry in conceptual design and offline software development [26][27][28].…”

Section: A Detector Description In Hep Softwarementioning

confidence: 99%

A method of detector description transformation to Unity and its application in BESIII

Huang,

Li,

Qian

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Detector and event visualization is an essential part in software of HEP experiments. Modern visualization technique and multimedia production platform, such as Unity, provides more fancy display effects and professional extensions for visualization in HEP experiments. In this work, a method of automatic detector description transformation is presented, which can convert the complicated HEP detector geometry from GDML in offline software to 3D modeling in Unity. The method has been successfully applied in the BESIII experiment, and can be further developed into applications such as event display, data monitoring or virtual reality. It has great potentials in detector design, offline software development, physics analysis and outreach for the next generation HEP experiments, as well as applications in nuclear techniques for industry.

show abstract

Improving the machine learning based vertex reconstruction for large liquid scintillator detectors with multiple types of PMTs

Cited by 2 publications

References 25 publications

Energy reconstruction for large liquid scintillator detectors with machine learning techniques: aggregated features approach

Energy reconstruction for large liquid scintillator detectors with machine learning techniques: aggregated features approach

A method of detector description transformation to Unity and its application in BESIII

Contact Info

Product

Resources

About