Searching for Neutrinoless Double-Beta Decay with XENONnT and AI

XENONnT employs a large target mass and dual-phase TPC to achieve unparalleled sensitivity in rare event searches. The neutrinoless double-beta decay searches at XENONnT encounter limitations due to gamma-rays emitted by the detector material. Therefore, a TextCNN (convolutional neural network for text) model with waveform augmentation is designed to extract maximum information from the detector data and therefore improve the background discrimination.

In this talk, we will show the TextCNN model development and performance, as well as the most recent progress on the 0𝜈𝛽𝛽 searches at XENONnT. The model demonstrates remarkable capability, achieving over 60% background rejection while maintaining a 90% signal acceptance. It significantly improved the background rejection for 0𝜈𝛽𝛽 searches at XENONnT, which can potentially improve the sensitivity of the 0𝜈𝛽𝛽 search for ¹³⁶Xe by over 30%. This highlights the potential of future dark matter experiments such as XLZD to achieve heightened sensitivity to 0𝜈𝛽𝛽 decay.