Machine-learning-enhanced event selection for BEGe detectors in the VIP experiment

The VIP collaboration operates a Broad-Energy Germanium (BEGe) detector at the Gran Sasso National Laboratory to measure radiation in the few-keV to 100 keV range, aiming to search for spontaneous collapse-induced radiation and atomic transitions violating the Pauli Exclusion Principle. Here, we present a machine-learning-based upgrade for the BEGe detector of an event-selection strategy aimed at improving the efficiency in detecting low-energy events down to 10 keV. The method employs a denoising autoencoder to suppress electronic and microphonic noises and reconstruct pulse shapes, followed by a convolutional neural network that classifies waveforms as normal single-site events or anomalous events. The workflow was validated on a dataset comprising more than 20,000 waveforms recorded in 2021. The classifier achieves a receiver operating characteristic (ROC) curve with an area under the curve (AUC) of 0.99 and an accuracy of 95%. Applying this procedure lowers the minimum detectable energy of the final spectrum to approximately 10 keV. It also yields a measurable enhancement in spectral quality, including an improvement of about 14% in the signal-to-background ratio and improvement of the energy resolution for the characteristic Pb and Bi gamma lines. These developments enhance the sensitivity of the BEGe detector to rare low-energy signals and provide a scalable framework for future precision tests of quantum foundations in low-background environments.