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

Abstract

Large-scale detectors consisting of a liquid scintillator target surrounded by an array of photo-multiplier tubes (PMTs) are widely used in the modern neutrino experiments: Borexino, KamLAND, Daya Bay, Double Chooz, RENO, and the 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 the JUNO detector, the most advanced 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 the inverse beta decay channel. We consider the following models: Boosted Decision Trees and Fully Connected Deep Neural Network, trained on aggregated features, calculated using the information collected by PMTs. We describe the details of our feature engineering procedure and show that machine learning models can provide the energy resolution $$\sigma = 3\%$$ 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.

Reactor Neutrino Energy Reconstruction with Machine Learning Techniques for the JUNO Experiment

“…Figure 3 shows the energy reconstruction performance (resolution and bias) of the BDT and FCDNN models. The hyperparameter optimization of the models is performed and follows the same procedure as in Ref [3]. In this study, we present an application of machine learning techniques for precise energy reconstruction for the JUNO detector and its satellite detector TAO.…”

“…The study is a continuation of Ref. [3] and uses an updated JUNO software [4] and also adds information collected by SPMTs. Moreover, the transferability of the approach to other LS-based detectors is described with the example of the JUNO-TAO detector (Section 2.2).…”

“…Table 1 collects and briefly describes the full set of engineered features and the detailed description can be found in Ref. [3]. AccumCharge is the total charge accumulated on all fired PMTs and it is at first order proportional to 𝐸 dep .…”

“…By using a greedy algorithm for feature selection described in Ref. [3], we kept the following 17 features (sorted by importance):…”

Neutron source-based event reconstruction in the JUNO detector

“…Other event reconstruction algorithms which have been developed for the reactor neutrino analysis in JUNO can be found elsewhere [3][4][5][6][7][8][9][10].…”