NOvA is a long-baseline neutrino experiment studying neutrino oscillations with Fermilab’s NuMI beam. A convolutional neural network (CNN) that analyzes topological features is used to determine neutrino flavor in both the near and far detectors and observe the disappearance of muon neutrinos and the appearance of electron neutrinos. Alternative approaches to flavor identification using machine learning are being investigated with the goal of developing a network trained with both event-level and particle-level images in addition to reconstructed physical variables while maintaining the performance of the CNN. Such a network could be used to analyze the individual prediction importances of these inputs. An original network that uses a combination of transformer and MobileNet CNN blocks will be discussed.