Improving CUORE Energy Reconstruction Using Machine Learning*

CUORE (Cryogenic Underground Observatory for Rare Events), located in the Laboratori Nazionali del Gran Sasso in Italy, is an experiment designed to search for neutrinoless double beta decay (0νββ) in ¹³⁰Te. Neutrinoless double beta decay is a theoretically predicted radioactive decay process that, if observed, would determine the Majorana nature of the neutrino. This would indicate that lepton number is not conserved and could help explain the matter antimatter asymmetry in the known universe. I present an analysis on the applications of different machine learning algorithms, including PCA (Principal Component Analysis) and linear regression, on improving energy reconstruction for CUORE. Energy resolution is of great importance for CUORE as it would help better distinguish 0νββ signals from background events. Currently, CUORE uses a software trigger algorithm called an Optimum Filter (OF) trigger for energy reconstruction. However, OF assumes that signal shape is independent of energy which isn’t true in CUORE.  A machine learning aided reconstruction could mitigate the energy dependence of pulse shapes and improve the overall energy resolution of the CUORE detector. I will present the current results of this analysis.