KamNet: An Integrated Spatiotemporal Deep Neural Network for Rare Event Search in KamLAND-Zen
This work addresses the challenge of improving rare event searches in particle physics detectors, representing an incremental advancement in applying geometric deep learning to spatiotemporal data.
The authors tackled the problem of rare event detection in KamLAND-Zen using a new deep learning model called KamNet, which outperformed conventional CNNs in simulations and increased sensitivity to neutrinoless double beta decay events.
Rare event searches allow us to search for new physics at energy scales inaccessible with other means by leveraging specialized large-mass detectors. Machine learning provides a new tool to maximize the information provided by these detectors. The information is sparse, which forces these algorithms to start from the lowest level data and exploit all symmetries in the detector to produce results. In this work we present KamNet which harnesses breakthroughs in geometric deep learning and spatiotemporal data analysis to maximize the physics reach of KamLAND-Zen, a kiloton scale spherical liquid scintillator detector searching for neutrinoless double beta decay ($0νββ$). Using a simplified background model for KamLAND we show that KamNet outperforms a conventional CNN on benchmarking MC simulations with an increasing level of robustness. Using simulated data, we then demonstrate KamNet's ability to increase KamLAND-Zen's sensitivity to $0νββ$ and $0νββ$ to excited states. A key component of this work is the addition of an attention mechanism to elucidate the underlying physics KamNet is using for the background rejection.