Improvement of machine learning-based vertex reconstruction for large liquid scintillator detectors with multiple types of PMTs

Machine learning in nuclear physics at low and intermediate energies

“…The models are trained on Monte Carlo simulation data. The vertex coordinates is around σ x,y,z = 10 cm at E vis = 1 MeV and decreases at higher energies, and the energy resolution is around σ E = 3% at E vis = 1 MeV, which resolution satisfies the requirements posed by the JUNO experiment [145,146]. In PandaX experiment [147], the performance of CNN in double beta decay events classification is much better than topological method [148].…”

Phase Transition Study Meets Machine Learning

“…One way to generally characterize dense nuclear matter is using nuclear empirical parameters defined from a series expansion around the isospin asymmetry. [23,[51][52][53] Among them, nuclear symmetry energy is critical in encoding the energy cost to make the matter more neutron rich. It is thus intriguing to obtain effective constraints over the nuclear symmetry energy with measurements in HICs.…”

“…This short review aims to elucidate the synergies between ML [22][23][24][25][26][27][28] and phase transition studies, and how these interdisciplinary collaborations are paving the way for new insights in physics. We delve into recent break-throughs where ML methods, including supervised and unsupervised learning, have been applied to phase transition studies of nuclear matter, enabling a finer characterization of the QCD phase diagram and a deeper understanding of the nuclear matter properties.…”

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

“…In the subsequent paper [14], 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 one, based on the PMT-wise gained information. On the other hand, vertex reconstruction requires granular information both with traditional and ML algorithms, see [15,16]. The actual research aims to further investigate the potential of the aggregated feature approach and to study two models: Boosted Decision Trees and Fully Connected Deep Neural Network for energy reconstruction in JUNO.…”