Regression CNNs for Reconstructing Neutrino Energy in the NOvA Experiment

NOvA is an accelerator neutrino experiment with an 810 km baseline. Using the NuMI beam located at Fermilab, NOvA measures electron neutrino appearance and muon neutrino disappearance at its far detector in northern Minnesota. NOvA uses deep learning in many different aspects of its reconstruction process. In this talk, we focus on convolutional neural networks (CNN) used to reconstruct energy. These regression CNNs take the raw cells from the detector as input, and give an estimate of the neutrino energy, or the energy of part of the neutrino event. Regression CNNs have been applied to both electron neutrino and muon neutrino events. The performance of these CNN-based methods sees an improvement compared to the traditional method.