Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Transforming jet flavour tagging at ATLAS

Aad, G. ; Åkesson, T.P.A. LU orcid ; Astrand, K.S.V. LU ; Calic, L. LU orcid ; Doglioni, C. LU ; Ekman, P.A. LU orcid ; Hedberg, V. LU ; Herde, H. LU orcid ; Konya, B. LU and Lytken, E. LU orcid , et al. (2026) In Nature Communications 17(1).
Abstract
Jet flavour tagging enables the identification of jets originating from heavy-flavour quarks in proton–proton collisions at the Large Hadron Collider, playing a critical role in its physics programmes. This paper presents GN2, a transformer-based flavour tagging algorithm deployed by the ATLAS Collaboration that represents a different methodology compared to previous approaches. Designed to classify jets based on the flavour of their constituent particles, GN2 processes low-level tracking information in an end-to-end architecture and incorporates physics-informed auxiliary training objectives to enhance both interpretability and performance. Its performance is validated in both simulation and collision data. The measured c-jet (light-jet)... (More)
Jet flavour tagging enables the identification of jets originating from heavy-flavour quarks in proton–proton collisions at the Large Hadron Collider, playing a critical role in its physics programmes. This paper presents GN2, a transformer-based flavour tagging algorithm deployed by the ATLAS Collaboration that represents a different methodology compared to previous approaches. Designed to classify jets based on the flavour of their constituent particles, GN2 processes low-level tracking information in an end-to-end architecture and incorporates physics-informed auxiliary training objectives to enhance both interpretability and performance. Its performance is validated in both simulation and collision data. The measured c-jet (light-jet) rejection in data is improved by a factor of 3.5 (1.8) for a 70% b-jet tagging efficiency, compared to the previous algorithm. GN2 provides substantial benefits for physics analyses involving heavy-flavour jets, such as measurements of Higgs boson pair production and the couplings of bottom and charm quarks to the Higgs boson, and demonstrates the impact of advanced machine learning methods in experimental particle physics. © The Author(s) 2025. (Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; ; ; ; and , et al. (More)
; ; ; ; ; ; ; ; ; ; ; ; ; and (Less)
author collaboration
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
algorithm, experimental study, jet, machine learning, tagging, training, article, controlled study, flavor
in
Nature Communications
volume
17
issue
1
article number
541
publisher
Nature Publishing Group
external identifiers
  • scopus:105027769260
  • pmid:41535252
ISSN
2041-1723
DOI
10.1038/s41467-025-65059-6
language
English
LU publication?
yes
id
9be2f545-d930-4264-bacc-e8b3846eb9e7
date added to LUP
2026-04-10 09:59:49
date last changed
2026-04-11 03:36:20
@article{9be2f545-d930-4264-bacc-e8b3846eb9e7,
  abstract     = {{Jet flavour tagging enables the identification of jets originating from heavy-flavour quarks in proton–proton collisions at the Large Hadron Collider, playing a critical role in its physics programmes. This paper presents GN2, a transformer-based flavour tagging algorithm deployed by the ATLAS Collaboration that represents a different methodology compared to previous approaches. Designed to classify jets based on the flavour of their constituent particles, GN2 processes low-level tracking information in an end-to-end architecture and incorporates physics-informed auxiliary training objectives to enhance both interpretability and performance. Its performance is validated in both simulation and collision data. The measured c-jet (light-jet) rejection in data is improved by a factor of 3.5 (1.8) for a 70% b-jet tagging efficiency, compared to the previous algorithm. GN2 provides substantial benefits for physics analyses involving heavy-flavour jets, such as measurements of Higgs boson pair production and the couplings of bottom and charm quarks to the Higgs boson, and demonstrates the impact of advanced machine learning methods in experimental particle physics. © The Author(s) 2025.}},
  author       = {{Aad, G. and Åkesson, T.P.A. and Astrand, K.S.V. and Calic, L. and Doglioni, C. and Ekman, P.A. and Hedberg, V. and Herde, H. and Konya, B. and Lytken, E. and Poettgen, R. and Smirnova, O. and Sur, N. and Wallin, E.J. and Zwalinski, L.}},
  issn         = {{2041-1723}},
  keywords     = {{algorithm; experimental study; jet; machine learning; tagging; training; article; controlled study; flavor}},
  language     = {{eng}},
  number       = {{1}},
  publisher    = {{Nature Publishing Group}},
  series       = {{Nature Communications}},
  title        = {{Transforming jet flavour tagging at ATLAS}},
  url          = {{http://dx.doi.org/10.1038/s41467-025-65059-6}},
  doi          = {{10.1038/s41467-025-65059-6}},
  volume       = {{17}},
  year         = {{2026}},
}