Machine Learning Approaches to Event Reconstruction for CRES

• Drew Byron (University of Washington)

Room: B205 The $^6$He-CRES experiment at the University of Washington CENPA aims to precisely measure the Fierz coefficient $b_{fierz}$ which parameterizes exotic currents in the weak interaction representing a violation of SM physics. A measurement of $b_{fierz}$ with a $10^{-3}$ uncertainty would be competitive with current LHC searches for tensor currents. We use Cyclotron Radiation Emission Spectroscopy (CRES), a technique first demonstrated by the Project 8 collaboration aiming at a determination of the neutrino mass via a measurement of the $^3{\rm H}$ beta spectrum. The technique determines the beta energy by measuring the frequency of the cyclotron radiation of betas in a magnetic field. CRES events are radio frequency chirps with various energy dependent properties. The accurate determination of signal parameters such as starting frequency, slope, and signal-to-noise ratio over a wide range in energies (from keV to MeV) is crucial to a high precision measurement of beta spectra. A brief experimental overview, the structure of our signals, and two machine learning based approaches for event reconstruction will be presented.