Machine-Learning Applications in Baikal-GVD: Current Status

Abstract

In this report we present machine-learning-based approaches for analyzing Baikal-GVD data. The framework addresses five key challenges in neutrino detection: suppression of air-shower-induced events, rejecting noise activations of optical modules, classification of track and cascade-like hits, reconstruction of neutrino incoming angles, and energy estimation. For each task, we discuss the physical motivation and demonstrate the performance metrics. We introduce a data processing pipeline that incorporates these neural networks and discuss how it can improve both the accuracy and efficiency of data analysis in the Baikal-GVD experiment.