Performance versus resilience in modern quark-gluon tagging
Discriminating quark-like from gluon-like jets is, in many ways, a key challenge for many LHC analyses. First, we use a known difference in PYTHIA and HERWIG simulations to show how decorrelated taggers would break down when the most distinctive feature is aligned with theory uncertainties. We propo...
| Hovedforfatter: | |
|---|---|
| Format: | Article |
| Sprog: | English |
| Udgivet: |
SciPost
2023-12-01
|
| Serier: | SciPost Physics Core |
| Online adgang: | https://scipost.org/SciPostPhysCore.6.4.085 |
| _version_ | 1827586495813255168 |
|---|---|
| author | Anja Butter, Barry M. Dillon, Tilman Plehn, Lorenz Vogel |
| author_facet | Anja Butter, Barry M. Dillon, Tilman Plehn, Lorenz Vogel |
| author_sort | Anja Butter, Barry M. Dillon, Tilman Plehn, Lorenz Vogel |
| collection | DOAJ |
| description | Discriminating quark-like from gluon-like jets is, in many ways, a key challenge for many LHC analyses. First, we use a known difference in PYTHIA and HERWIG simulations to show how decorrelated taggers would break down when the most distinctive feature is aligned with theory uncertainties. We propose conditional training on interpolated samples, combined with a controlled Bayesian network, as a more resilient framework. The interpolation parameter can be used to optimize the training evaluated on a calibration dataset, and to test the stability of this optimization. The interpolated training might also be useful to track generalization errors when training networks on simulations. |
| first_indexed | 2024-03-09T00:03:25Z |
| format | Article |
| id | doaj.art-d8aaa8bee95846b5ac7585d6e655f1d2 |
| institution | Directory Open Access Journal |
| issn | 2666-9366 |
| language | English |
| last_indexed | 2024-03-09T00:03:25Z |
| publishDate | 2023-12-01 |
| publisher | SciPost |
| record_format | Article |
| series | SciPost Physics Core |
| spelling | doaj.art-d8aaa8bee95846b5ac7585d6e655f1d22023-12-12T15:41:52ZengSciPostSciPost Physics Core2666-93662023-12-016408510.21468/SciPostPhysCore.6.4.085Performance versus resilience in modern quark-gluon taggingAnja Butter, Barry M. Dillon, Tilman Plehn, Lorenz VogelDiscriminating quark-like from gluon-like jets is, in many ways, a key challenge for many LHC analyses. First, we use a known difference in PYTHIA and HERWIG simulations to show how decorrelated taggers would break down when the most distinctive feature is aligned with theory uncertainties. We propose conditional training on interpolated samples, combined with a controlled Bayesian network, as a more resilient framework. The interpolation parameter can be used to optimize the training evaluated on a calibration dataset, and to test the stability of this optimization. The interpolated training might also be useful to track generalization errors when training networks on simulations.https://scipost.org/SciPostPhysCore.6.4.085 |
| spellingShingle | Anja Butter, Barry M. Dillon, Tilman Plehn, Lorenz Vogel Performance versus resilience in modern quark-gluon tagging SciPost Physics Core |
| title | Performance versus resilience in modern quark-gluon tagging |
| title_full | Performance versus resilience in modern quark-gluon tagging |
| title_fullStr | Performance versus resilience in modern quark-gluon tagging |
| title_full_unstemmed | Performance versus resilience in modern quark-gluon tagging |
| title_short | Performance versus resilience in modern quark-gluon tagging |
| title_sort | performance versus resilience in modern quark gluon tagging |
| url | https://scipost.org/SciPostPhysCore.6.4.085 |
| work_keys_str_mv | AT anjabutterbarrymdillontilmanplehnlorenzvogel performanceversusresilienceinmodernquarkgluontagging |