Performance of top-quark and $W$-boson tagging with ATLAS in Run 2 of the LHC

TL;DR

This paper evaluates and compares various advanced jet tagging techniques for identifying top quarks and W bosons in LHC Run 2 data, including cut-based, multivariate, and deep neural network methods.

Contribution

It introduces optimized taggers using jet substructure observables and deep learning, enhancing identification performance for highly boosted particles.

Findings

Deep neural networks improve tagging accuracy.

Multivariate taggers outperform traditional methods.

Optimized taggers are validated with real LHC data.

Abstract

The performance of taggers for hadronically decaying top quarks and $W$ bosons in $pp$ collisions at $\sqrt{s}$ = 13 TeV recorded by the ATLAS experiment at the Large Hadron Collider is presented. A set of techniques based on jet shape observables are studied to determine a set of optimal cut-based taggers for use in physics analyses. The studies are extended to assess the utility of combinations of substructure observables as a multivariate tagger using boosted decision trees or deep neural networks in comparison with taggers based on two-variable combinations. In addition, for highly boosted top-quark tagging, a deep neural network based on jet constituent inputs as well as a re-optimisation of the shower deconstruction technique is presented. The performance of these taggers is studied in data collected during 2015 and 2016 corresponding to 36.1 fb$^{-1}$ for the $t\bar{t}$ and $\gamma+$jet and 36.7 fb$^{-1}$ for the dijet event topologies.