Speaker
Description
IceCube-Gen2 is the planned high-energy extension of the current IceCube detector, featuring an optical array nearly eight times larger than that of IceCube and composed of novel optical modules containing multiple photomultiplier tubes (PMTs), providing enhanced and nearly omnidirectional sensitivity. This configuration is expected to significantly increase event statistics and improve the discovery potential for astrophysical neutrino sources. Fully exploiting these advances requires new reconstruction and event selection techniques adapted to the detector geometry and sensor design. In this contribution, we present ongoing work on deep learning-based methods for neutrino event direction and energy reconstruction in the IceCube-Gen2 Optical Array, together with classifiers designed to suppress atmospheric muon bundle background and identify starting events. We discuss the performance of these methods and their potential impact on future point-source searches.
