Dortmund 2021 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 42: Neutrino astronomy II

T 42.4: Talk

Tuesday, March 16, 2021, 16:45–17:00, Tq

Redesigning the IceCube processing pipeline with deep learning — •Christian Haack and Theo Glauch for the IceCube collaboration — TU München

The IceCube Neutrino Observatory is a cubic-kilometer sized neutrino detector located at the geographic South Pole. The main background for neutrino searches are atmospheric muons, induced by cosmic-ray air showers. IceCube employs multiple levels of software filters, involving various event reconstructions, to reduce the atmospheric muon background and classify the events based on their topology. While this pipeline has enabled many scientific breakthroughs, such as the measurement of the diffuse astrophysical neutrino flux and the detection of the first high-energy neutrino point source, many of its components have been in place for more than ten years. This makes the filtering pipepline increasingly challenging to maintain. Additionally, detector extensions, such as the IceCube Upgrade, are difficult to incorporate in the existing filtering systems. However, recent developments in deep-learning based filtering and reconstruction algorithms have the potential to drastically simplify the filtering pipeline. In this talk, we will outline how a deep-learning based filtering system for IceCube might look like.