Total: 1
Atmospheric neutrino oscillations are important to the study of neutrino properties, including the neutrino mass ordering problem. A good capability to identify neutrinos' flavor and neutrinos against antineutrinos is crucial in such measurements. In this paper, we present a machine-learning-based approach for identifying atmospheric neutrino events in a large homogeneous liquid scintillator detector. This method extracts features that reflect event topologies from PMT waveforms and uses them as input to machine learning models. In addition, neutron-capture information is utilized to achieve neutrino versus antineutrino discrimination. Preliminary performances based on Monte Carlo simulations are presented, which demonstrate such detector's potential in the future measurements of atmospheric neutrinos.