D. Gooding: Enhancing the Neutrinoless Double Beta Decay Sensitivity of SNO+ with Deep Learning

Measurements of neutrino oscillations have confirmed that neutrinos are massive, but other questions remain: What are the absolute neutrino masses? Which neutrino is the lightest? Answers to these questions don't only provide all of the remaining unknown quantities in The Standard Model. They also have consequences for our understanding of how fundamental particles acquire mass. One way to investigate neutrino mass is to determine whether or not the neutrino is its own anti-particle, or "Majorana" in nature. This can be done experimentally by measuring neutrinoless double-beta decay. Once the SNO+ experiment is loaded with 130-Te, it will join several other liquid scintillator detectors currently looking for this process. This talk describes a log-likelihood approach to quantifying current background levels in the SNO+ detector, as well as a toy Monte-Carlo study to calculate the experiment's projected sensitivity to neutrinoless double beta decay after 5.25e-02 ton yrs of exposure. This work also describes the additional background rejection achieved by KamNet, A novel deep neural network developed by the KamLAND collaboration. At 90% confidence, the deployment of this tool increases the projected neutrinoless double beta decay half-life limit of SNO+ by 41.3%, from T > 5.2e+25 years to T > 7.3e+25 years.