Search for vector-like quark pairs with DNN jet identification in single lepton events

We present a search for vector-like quark pairs using deep neural network (DNN) identification methods for boosted particle jets. We search for vector-like T quarks in √s = 13 TeV proton-proton collision data collected by the CMS experiment during 2017-2018 with one charged electron or muon, missing transverse energy, and several large radius jets. The DeepAK8 identification algorithm is used to identify jets from the T quark decays, that are reconstructed from at least 3 jets and the leptons. An additional DNN is trained on events that are not categorized as a specific TT decay to separate signal from background. Compared to similar CMS searches using data collected in 2016, the sensitivity of this search has significantly increased and the new method provides more information about the mass and decay modes of the TT pairs.