Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 43: Data, AI, Computing 3 (pointclouds & graphs)

T 43.8: Vortrag

Dienstag, 5. März 2024, 17:45–18:00, Geb. 30.33: MTI

Enhancing Neutrino Event Classification in the IceCube Observatory Using a Neural-Network Approach — •Philipp Soldin, Jakob Böttcher, Philipp Fürst, Erik Ganster, Michael Handt, Johanna Hermannsgabner, and Christopher Wiebusch — RWTH Aachen University

The IceCube Neutrino Observatory is a particle detector located at the geographic south pole. It is a cubic kilometer in size and detects neutrinos by measuring the Cherenkov light from their interaction products. One of the main challenges in IceCube is accurately classifying neutrino events based on these measured signals. Previous attempts achieved high accuracy but had to aggregate large amounts of data for processing. However, new deep learning techniques, such as transformer and graph-based architectures, allow for the use of more signal data without prior aggregation. This pure signal data enables the utilization of intricate signal details and improves the selection efficiency. The talk presents the latest advances in this approach and its results.

Keywords: IceCube; Neural Networks