Machine Learning based background correction for jet shapes in Pb-Pb collisions at 5.02 TeV

Jet shapes and, furthermore, jet substructure observables are of utmost interest for the field of heavy-ion physics. However, the overwhelmingly large background from soft processes complicates a measurement in particular for low jet transverse momenta: Both, the jet energy scale as well as the jet shape itself, are strongly affected by the background. While promising studies on the correction of the jet energy scale and first pilot studies on two-parameter regression (jet momentum/jet shape) were already made, detailed studies of the latter are yet missing. In this project, we evaluate the performance of several machine learning algorithms and combinations of input parameters on the background correction of the jet properties.