The identification of high transverse momentum top quarks is an essential ingredient of many searches for new physics. This thesis presents, for the first time, the performance of shower deconstruction, a new top-tagging algorithm, using data collected with the ATLAS detector at the LHC. The distribution of the shower deconstruction observable, the likelihood ratio χSD, is shown to be very well modelled by modern Monte Carlo generators, and is compared to several other algorithms and jet substructure observables. Shower deconstruction is shown to have the best quark- and gluon-jet background rejection over a wide range of signal efficiency. Vector-like quarks are a feature of several new physics models. As they do not gain their mass from a Yukawa coupling, they are still allowed by constraints set by Higgs production cross section measurements. These, so-called, fermionic-partners can play a role similar to that of bosonic-partners in supersymmetric models, providing a solution for the naturalness problem. This thesis presents the first dedicated search for single production of vector-like quarks, using data collected with the ATLAS detector, introducing novel strategies for separating signal from background. This work extends the exclusion limits from production of pairs of vector-like quarks, from 700 GeV to 950 GeV, for suitably large vector-like quarks couplings. Also, exclusion limits on the coupling versus vector-like quark mass space are presented for the first time. Large-R jet triggers are first introduced in ATLAS during Run 1 to cover topologies where these jets play a predominant role. In this thesis, several performance metrics of these triggers are presented, using data collected with the ATLAS detector, highlighting their advantages over standard jet triggers. Also, an example of how these triggers can be combined to lepton triggers to increase the efficiency of a search for new physics is shown.