Machine Learning-Based Reconstruction for Precision Neutrino Oscillation Measurements in NOvA

The NOvA experiment studies neutrino oscillations using two detectors positioned off-axis from the NuMI beam at Fermilab. Building on its recent three-flavor oscillation results, which achieved the most precise measurement of the atmospheric mass-splitting to date, NOvA continues to refine reconstruction and analysis techniques to maximize physics sensitivity. Machine learning (ML) methods play an increasingly important role throughout NOvA's reconstruction chain, including event classification, particle identification, vertex and energy reconstruction. These approaches have improved both efficiency and purity compared to traditional techniques and have enabled new analysis strategies as the experiment's dataset continues to grow. Ongoing developments focus on refining existing models and exploring emerging architectures to further enhance reconstruction performance and physics reach.