Speaker
Description
Many modern astroparticle physics experiments rely on simulation of air showers to model their signal and background. The challenge with these simulations is the accurate modeling of large amounts of particle interactions in realistic media. The current solution is time- and resource-intensive Monte Carlo campaigns tuned to the specific experiments. For the IceCube Neutrino Observatory, this becomes especially challenging in the Southern Sky where muons produced in air showers represent the dominant background at lower Energies. We explore the use of generative AI to generate particle showers as an alternative to classical full Monte Carlo generation and highlight how it can be used to build improved event selections in regions of the sky where modeling of the air shower-induced background is critical.