Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Classifying hadronic objects in ATLAS with ML/AI algorithms

Leonardo Toffolin

Abstract

The identification of hadronic final states plays a crucial role in the physics programme of the ATLAS Experiment at the CERN LHC. Sophisticated artificial intelligence (AI) algorithms are employed to classify jets according to their origin, distinguishing between quark- and gluon-initiated jets, and identifying hadronically decaying heavy objects such as W bosons and top quarks. This contribution summarises recent developments in constituent-based tagging architectures, including graph neural networks (GNNs) and transformer-based approaches, their performance in simulated and real data, and future perspectives towards data-driven optimisation and model-independent tagging strategies.

Classifying hadronic objects in ATLAS with ML/AI algorithms

Abstract

The identification of hadronic final states plays a crucial role in the physics programme of the ATLAS Experiment at the CERN LHC. Sophisticated artificial intelligence (AI) algorithms are employed to classify jets according to their origin, distinguishing between quark- and gluon-initiated jets, and identifying hadronically decaying heavy objects such as W bosons and top quarks. This contribution summarises recent developments in constituent-based tagging architectures, including graph neural networks (GNNs) and transformer-based approaches, their performance in simulated and real data, and future perspectives towards data-driven optimisation and model-independent tagging strategies.
Paper Structure (4 sections, 4 figures)

This paper contains 4 sections, 4 figures.

Table of Contents

  1. Machine Learning for jet tagging
  2. Quark/gluon and flavour tagging
  3. $W$-boson and top-quark tagging
  4. Summary and Outlook

Figures (4)

  • Figure 1: Architecture of the DeParT algorithm DeParT.
  • Figure 2: Performance of the DeParT transformer-based algorithm for quark/gluon tagging compared with FC DNN, PFN, EFN, and ParticleNet architectures, expressed in terms of gluon-jet rejection $\epsilon^{-1}_{g}$. The rejection is shown as a function of (a) quark identification efficiency $\epsilon_{q}$ and (b) $p_{\rm T}$. Taken from DeParT.
  • Figure 3: Performance of the ParT algorithm as a function of the jet identification efficiency $\epsilon_{\rm sig}$ (a) and jet $p_{\rm T}$ (b), compared with EFN, PFN and ParticleNet. Taken from W tagging with constituents.
  • Figure 4: (a) Background rejection performance of the LundNet and LundNetANN taggers for $W$-boson identification. The LundNetANN algorithm, despite a smaller rejection power, has the ability of decorrelating the performance from jet mass and reproducing better the QCD jets contribution, as shown in (b). Taken from W tagging LJP.