Huw Haigh

24 November 2021
ACAT 2021

Abstract: The search for new particles requires robust classification methods, with the ability to be optimised for unknown cross sections and particle masses. We present a new loss function, 'Punzi-loss', based on the so-called Punzi figure of merit (FOM). We refer to a neural network trained with the Punzi-loss function as a 'Punzi-net', and investigate its application to the search for invisible decays of the hypothetical Z' boson produced in the process e+e- -> mu+mu-Z' at the Belle II experiment.