Particle Identification in Highly Segmented Detector SuperFGD of the T2K Neutrino Experiment

A deep neural network with long-short-term memory and permutation-invariant feature encoder is proposed for distinguishing electrons, photons, muons, pions, and protons in the novel detector SuperFGD. SuperFGD is a highly granular scintillator target that is a central component of the upgraded ND280 near detector in the T2K neutrino experiment. Precise particle identification is essential for reconstructing neutrino interactions and reducing systematic uncertainties in T2K’s oscillation measurements. The proposed model demonstrates robust discrimination performance across a range of particle types.