Speaker
Description
Neutrinos in dense astrophysical environments such as core-collapse supernovae (CCSNe) and neutron star mergers (NSM) can undergo FFCs, which could develop on extremely small scales. A necessary condition for the occurrence of FFCs is the presence of a zero crossing in the electron lepton number (ELN) angular distribution of neutrinos. In this work, we explore machine learning (ML) approaches to detect non-axisymmetric ELN crossings in these environments.
While the ML models achieve good overall performance, their accuracies vary across different test datasets, reflecting the influence of environment-dependent features on the ML performance. When applied to already flavor-equilibrated ELN angular distributions, the performance of our ML model is comparatively lower, owing to the absence of heavy-lepton flavor information in the training inputs. However, when the crossing definition is restricted to the same flavor information used during training, the model performance improves significantly, demonstrating that the ML models remain robust when the test data are consistent with the training feature space.