Identification of hadronic tau lepton decays using a deep neural network

Abstract A new algorithm is presented to discriminate reconstructed hadronic decays of tau leptons ( τ h ) that originate from genuine in the CMS detector against candidates quark or gluon jets, electrons, muons. The inputs information all particles vicinity a candidate and employs deep neural network with convolutional layers efficiently process inputs. This leads significantly improved performance compared previously used one. For example, efficiency for pass discriminator jets increases by 10–30% given jets. Furthermore, more efficient reconstruction introduced incorporates additional decay modes. superior muons method are validated LHC proton-proton collision data at √ s = 13 TeV.