Speaker
Description
Missing transverse momentum (MET) is a critical observable for physics searches in proton-proton collisions at the Large Hadron Collider. This talk describes these various novel approaches and their performance. ATLAS employs a suite of working points for missing transverse momentum (MET) reconstruction, and each is optimal for different event topologies. A new neural network can exploit various event properties to pick the optimal working point on an event-by-event basis and also combine complementary information from each of the working points. The resulting regressed "METNet" offers improved resolution and pileup resistance across a number of different topologies compared to the current MET working points. Additionally, image-based de-noising neural network techniques are studied; these also provide significant resolution improvements and pileup resistance.