Data Mining Ice Cubes
share
IceCube is a 1 km3 scale neutrino telescope located at the geographic South Pole. The large number of reconstructed attributes as well as the small signal to background ratio in an atmospheric neutrino analysis makes IceCube well suited for a detailed study within the scope of machine learning. A systematic study intended to improve the event selection was carried out including a detailed feature selection using MRMR as well as the training and testing of a Random Forest. Finally, the forest was applied on IceCube data. A good agreement between data and Monte Carlo expectations has been observed.
READ FULL TEXT