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 poster 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). KamNet independently identified and vetoed periods with high background noise, leading to the first search for the Majorana nature of neutrinos in the inverted mass ordering region with KamLAND-Zen 800. Furthermore, KamNet has the potential to further improve KamLAND-Zen 800 limit if directly applied to the entire dataset. 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.