The ARIANNA experiment is an array of low powered radio detectors located in Antarctica. The aim of the detector is to study high-energy astrophysical phenomena with cosmic neutrino messengers. The extremely low flux of these neutrinos makes detector optimization crucial for gathering enough neutrino data. Therefore, the goal of this work was to create a real-time deep learning filter to increase our ability to measure neutrinos without changing the rigid data transmission rate. I will discuss the design, installation, and lab testing of a real-time deep learning filter in the current ARIANNA electronics. This filter increases the sensitivity of the detector by up to a factor of two. In addition, I will discuss an offline deep learning study using experimental and simulated data to determine how well experimental background data can be rejected. In both projects, a deep learning approach is found to give significantly better results compared to more traditional analysis techniques.
Join from PC, Mac, Linux, iOS or Android: https://stanford.zoom.us/j/98973156241?pwd=cEU5RFdlVXoyc0JTeTlDMkozKzQ5UT09
Peter Rowson, Federico Bianchini, Yifan Chen
(rowson@slac, fbianc@slac, cyifan@slac)
