Speaker
Description
The IceCube neutrino observatory is searching for point sources in the astrophysical neutrino flux. Relativistic muons created by muon-neutrinos offer a good angular resolution and are thus an ideal channel for the detection of points sources.
Recurrent neural networks (RNNs) are a class of artificial neural networks that capture the dynamics of sequential data by recurrently applying the network to each elements in a sequence. They retain a state from previous elements of the sequence and are thus able to aggregate information from arbitrarily long sequences. This makes RNNs well suited for time series data such as the signatures created by particles traveling through IceCube.
In this contribution I present a status report on directional reconstruction of muons in IceCube using recurrent neural networks.
