Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Dijet Resonance Search with Weak Supervision Using s =13 TeV pp Collisions in the ATLAS Detector

Aad, G ; Åkesson, Torsten LU orcid ; Bocchetta, Simona LU ; Corrigan, Eric Edward LU ; Doglioni, Caterina LU ; Geisen, Jannik LU orcid ; Gregersen, Kristian LU ; Brottmann Hansen, Eva LU ; Hedberg, Vincent LU and Jarlskog, Göran LU , et al. (2020) In Physical Review Letters 125(13).
Abstract
This Letter describes a search for narrowly resonant new physics using a machine-learning anomaly detection procedure that does not rely on signal simulations for developing the analysis selection. Weakly supervised learning is used to train classifiers directly on data to enhance potential signals. The targeted topology is dijet events and the features used for machine learning are the masses of the two jets. The resulting analysis is essentially a three-dimensional search A→BC, for mA∼O(TeV), mB,mC∼O(100 GeV) and B, C are reconstructed as large-radius jets, without paying a penalty associated with a large trials factor in the scan of the masses of the two jets. The full run 2 s=13 TeV pp collision dataset of 139 fb-1 recorded by the... (More)
This Letter describes a search for narrowly resonant new physics using a machine-learning anomaly detection procedure that does not rely on signal simulations for developing the analysis selection. Weakly supervised learning is used to train classifiers directly on data to enhance potential signals. The targeted topology is dijet events and the features used for machine learning are the masses of the two jets. The resulting analysis is essentially a three-dimensional search A→BC, for mA∼O(TeV), mB,mC∼O(100 GeV) and B, C are reconstructed as large-radius jets, without paying a penalty associated with a large trials factor in the scan of the masses of the two jets. The full run 2 s=13 TeV pp collision dataset of 139 fb-1 recorded by the ATLAS detector at the Large Hadron Collider is used for the search. There is no significant evidence of a localized excess in the dijet invariant mass spectrum between 1.8 and 8.2 TeV. Cross-section limits for narrow-width A, B, and C particles vary with mA, mB, and mC. For example, when mA=3 TeV and mBâ200 GeV, a production cross section between 1 and 5 fb is excluded at 95% confidence level, depending on mC. For certain masses, these limits are up to 10 times more sensitive than those obtained by the inclusive dijet search. These results are complementary to the dedicated searches for the case that B and C are standard model bosons. © 2020 CERN. (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
Anomaly detection, Germanium compounds, Large dataset, Machine learning, Mass spectrometry, Turing machines, Confidence levels, Invariant-mass spectra, Large Hadron Collider, Potential signal, Production cross section, Resonance searches, Signal simulation, Weakly supervised learning, Boron compounds, article, body weight, boson, hadron, human, mass spectrometry, punishment
in
Physical Review Letters
volume
125
issue
13
article number
131801
publisher
American Physical Society
external identifiers
  • scopus:85092801738
  • pmid:33034503
ISSN
1079-7114
DOI
10.1103/PhysRevLett.125.131801
language
English
LU publication?
yes
id
1b0c11e3-7c8a-4e8a-b2d1-f88481eb9624
date added to LUP
2020-11-10 08:39:57
date last changed
2023-04-10 22:59:32
@article{1b0c11e3-7c8a-4e8a-b2d1-f88481eb9624,
  abstract     = {{This Letter describes a search for narrowly resonant new physics using a machine-learning anomaly detection procedure that does not rely on signal simulations for developing the analysis selection. Weakly supervised learning is used to train classifiers directly on data to enhance potential signals. The targeted topology is dijet events and the features used for machine learning are the masses of the two jets. The resulting analysis is essentially a three-dimensional search A→BC, for mA∼O(TeV), mB,mC∼O(100 GeV) and B, C are reconstructed as large-radius jets, without paying a penalty associated with a large trials factor in the scan of the masses of the two jets. The full run 2 s=13 TeV pp collision dataset of 139 fb-1 recorded by the ATLAS detector at the Large Hadron Collider is used for the search. There is no significant evidence of a localized excess in the dijet invariant mass spectrum between 1.8 and 8.2 TeV. Cross-section limits for narrow-width A, B, and C particles vary with mA, mB, and mC. For example, when mA=3 TeV and mBâ200 GeV, a production cross section between 1 and 5 fb is excluded at 95% confidence level, depending on mC. For certain masses, these limits are up to 10 times more sensitive than those obtained by the inclusive dijet search. These results are complementary to the dedicated searches for the case that B and C are standard model bosons. © 2020 CERN.}},
  author       = {{Aad, G and Åkesson, Torsten and Bocchetta, Simona and Corrigan, Eric Edward and Doglioni, Caterina and Geisen, Jannik and Gregersen, Kristian and Brottmann Hansen, Eva and Hedberg, Vincent and Jarlskog, Göran and Kellermann, Edgar and Konya, Balazs and Lytken, Else and Mankinen, Katja and Marcon, Caterina and Mjörnmark, Ulf and Mullier, Geoffrey André Adrien and Pöttgen, Ruth and Poulsen, Trine and Skorda, Eleni and Smirnova, Oxana and Zwalinski, L}},
  issn         = {{1079-7114}},
  keywords     = {{Anomaly detection; Germanium compounds; Large dataset; Machine learning; Mass spectrometry; Turing machines; Confidence levels; Invariant-mass spectra; Large Hadron Collider; Potential signal; Production cross section; Resonance searches; Signal simulation; Weakly supervised learning; Boron compounds; article; body weight; boson; hadron; human; mass spectrometry; punishment}},
  language     = {{eng}},
  number       = {{13}},
  publisher    = {{American Physical Society}},
  series       = {{Physical Review Letters}},
  title        = {{Dijet Resonance Search with Weak Supervision Using s =13 TeV pp Collisions in the ATLAS Detector}},
  url          = {{http://dx.doi.org/10.1103/PhysRevLett.125.131801}},
  doi          = {{10.1103/PhysRevLett.125.131801}},
  volume       = {{125}},
  year         = {{2020}},
}