Sparse Convolutional Neural Network for NOvA

The NOvA (NuMI Off-axis v_e Appearance) experiment measures neutrino oscillations in a nearly pure muon neutrino beam over a 810 km baseline. NOvA has successfully used a Convolutional Neural Network as its main event selector and particle identifier for oscillation analysis since 2016. Despite having over 90% efficiency in classifying all neutrino types, our current network requires significant GPU resources during training and can only be applied to a window of activity around the beginning of each event. In order to reduce the computational cost, we have implemented a Sparse Convolutional Neural Network which performs convolutions only when the center pixel of the receptive field is non-zero. This reduces the size of training data and improves throughput. We have implemented a sparse FishNet architecture with PyTorch Lightning, a Python library providing high-level interface, which resulted in training accuracy of ~90%. We plan to utilize true neutrino energy information in the training in order to weight the accuracy at different energy regions to increase. In the future, we hope to incorporate semantic and instance segmentation simultaneously into the FishNet architecture for full end-to-end reconstruction.