Münster 2017 – wissenschaftliches Programm

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

T: Fachverband Teilchenphysik

T 23: Experimentelle Techniken der Astroteilchenphysik 2

T 23.7: Vortrag

Montag, 27. März 2017, 18:15–18:30, S 055

Neural Networks for Energy Reconstruction in the IceCube Neutrino Observatory — •Martin Brenzke, Jan Auffenberg, Christian Haack, René Reimann, and Christopher Wiebusch for the IceCube collaboration — III. Physikalisches Institut, RWTH Aachen University, D-52056 Aachen, Germany

Energy reconstruction of track-like events induced by muons is an essential part of the data analysis of the IceCube Neutrino Observatory. There already are sophisticated methods to reconstruct the energy of those events. However, the progress achieved in the recent decade in deep learning techniques makes them an interesting candidate for an alternative method for energy reconstruction, which might perform as well as or even better than the established algorithms. We focus on supervised learning techniques using recurrent neural networks and present first results of performance studies as well as comparisons to commonly used reconstruction methods.