Signal Classification Techniques for Searches and
Measurements at the LHC

Abstract

This thesis focuses on three different examples of techniques designed to extract
signal from background in the highly polluted by QCD environment of the proton-proton
collisions at the Large Hadron Collider. The first is an attempt at quark-gluon
tagging with the help of a simplified version of the shower deconstruction
approximation to the likelihood ratio. We find that it outperforms some frontrunners
in the field for a large variety of jet definitions and constraints, assuming
topocluster-like objects instead of hadrons as seeds. The second search is tasked
with identifying boosted W bosons, emitted from high virtuality quarks, thereby
measuring the effects of Sudakov logarithmic enhancement under different assumptions
of the systematic uncertainty. Finally, we examine the LHC's capability to
measure and constrain the strength of the ttH (H->bb) channel in an extensive search of
various modestly boosted phase space regions. Under optimistic assumptions about
the missing energy reconstruction in b-tagged jets and the handling of the systematic
uncertainty, we are able to exclude deviations on the order of 20% from the SM
expectation.