Machine Learning event identification in H → ττ analysis of CMS

• Michal Olszewski (University of Warsaw (PL))

Room: Room: B2.38 The H → τ+τ− channel (branching fraction about 6%) is currently one of two (next to H → bb̄ with branching fraction of about 57%) at the LHC best suited to perform measurements of 125 GeV Higgs boson couplings to fermions. The first observation of this process in a single experiment was achieved by the CMS Collaboration and published in 2017. The analysis summarized in the publication is driven by several characteristics specific for aforementioned Higgs boson decay mode. First, the tau lepton cannot by directly observed at the detector due to short lifetime. Second, it decays either leptonically or semi-leptonically due to invariant mass larger than mass of the lightest mesons. Furthermore, the large QCD background at LHC and presence of the neutrinos in the final state enforces usage of sophisticated algorithms for tau identification and ditau system mass reconstruction (respectively). In the first part of the talk I’ll shortly present how CMS analysis deals with above topics. In the second part I’ll show how machine learning techniques can by used to reinforce the analysis sensitivity to the signal. Spotkanie odbędzie się za pośrednictwem platformy zoom: Zoom meeting link: https://zoom.us/j/176261850 Zoom meeting ID: 176 261 850