Aachen 2019 – scientific programme

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

T: Fachverband Teilchenphysik

T 29: Deep Learning II

T 29.8: Talk

Tuesday, March 26, 2019, 17:45–18:00, H06

Application of Deep Neural Networks to Event Type Classification in IceCube — •Maximilian Kronmüller and Theo Glauch for the IceCube collaboration — Technical University of Munich

The IceCube Neutrino Telescope is able to measure an all-flavor neutrino flux in an energy range between 100 GeV and several PeV. Due to the different features of the neutrino interactions and the geometry of the detector all high-level analyses require a selection of suitable events as a first step. However, up to today, no algorithm exists that gives a generic prediction of an event's topology. One possible solution to this is the usage of deep neural networks, i.e. classification networks similar to the ones used in image recognition. The classifier that we present here is based on a modern InceptionResNet architecture and includes multi-task learning in order to broaden the field of application and increase the overall accuracy of the result. Despite a detailed discussion of the network's architecture we will also examine the performance and speed of the classifier for various tasks and possible applications in IceCube.