Speaker
Description
The Higgs boson production in association with a top quark pair plays a key role for studying the Yukawa coupling between the Higgs boson and the top quark. The coupling can be determined by measuring the cross-section of the $t\bar{t}H$ production in various final states using the $140\,\text{fb}^{-1}$ ATLAS dataset at $\sqrt{s}=13\,\text{TeV}$. Multi-lepton final states are rare but pure since most backgrounds are significantly suppressed.
The non-resonant $t\bar{t}H\to4\ell$ process has low production rate and has contributions from Higgs decay modes like $H\to WW^{*}$, $H\to\tau\tau$, and $H\to Z^{*}Z^{*}$. The dominant background arises from $t\bar{t}Z$, $ZZ$, and misidentified leptons from $t\bar{t}$ production. A multiclass dense neural network (DNN) is trained to separate signal events from these backgrounds and to define analysis regions. Additional fake lepton regions are defined to estimate the normalisation of the most important fake contributions. Partially unblinded fit results for the signal strength are presented for the $4\ell$ channel, but also the combined performance of other multi-lepton final states are shown.