Machine Learning Applied to Mult-Electron Events in Scintillator

Conversion electron spectroscopy is a viable tool when studying the nuclear phenomenon, shape coexistence. When a neutron-rich nucleus beta decays, a neutron transforms into a proton and emits an electron. Due to electromagnetic interactions, this can result in the ejection of an electron from the atom, a process called internal conversion. Because this process is essentially simultaneous in time, it is pivotal to differentiate between the electron emitted from the nucleus and the internal conversion electron emitted from the atom. Here we apply supervised machine learning algorithms to distinguish between one and two electron events, as well as determine the origin of the electron. With simulated data, we were able to successfully train a convolutional neural network (CNN) to distinguish between a one and two electron event with 96.79% accuracy. Furthermore, we successfully trained a CNN to predict the origin of the electron for one electron events. Our results show promise that our models' performance will generalize to experimental data. Once our models are complete, machine learning will be an important data analysis tool for conversion electron spectroscopy.