Cargando…

Jet flavour identification using Graph Neural Networks in CMS

<!--HTML--><p>The identification of the flavour of the parton that initiated the spray of stable particles forming a jet is&nbsp;a&nbsp;fundamental component of&nbsp;standard model measurements and searches for new physics with quarks and gluons in the final state. The first...

Descripción completa

Detalles Bibliográficos
Autor principal: Gouskos, Loukas
Lenguaje:eng
Publicado: 2022
Materias:
Acceso en línea:http://cds.cern.ch/record/2835621
_version_ 1780975650978398208
author Gouskos, Loukas
author_facet Gouskos, Loukas
author_sort Gouskos, Loukas
collection CERN
description <!--HTML--><p>The identification of the flavour of the parton that initiated the spray of stable particles forming a jet is&nbsp;a&nbsp;fundamental component of&nbsp;standard model measurements and searches for new physics with quarks and gluons in the final state. The first jet identification algorithms were developed using classical methods, however since the&nbsp;beginning of LHC Run 2 the development of&nbsp;significantly more advanced methods, making use of machine learning techniques, led to&nbsp;substantial improvement in&nbsp;performance, and hence in the physics reach.&nbsp;Particularly, Graph Neural Networks (GNNs) are emerging as an extremely powerful class of architectures;&nbsp;GNNs provide a much more natural representation for many&nbsp;tasks in High Energy Physics, including the task&nbsp;of jet identification.&nbsp;In today's seminar we will present a new generation of jet identification algorithms developed by CMS, which&nbsp;exploit&nbsp;customised&nbsp;GNNs tailored to the jet identification task, yielding a drastic improvement in performance&nbsp;compared to&nbsp;previous algorithms.&nbsp;</p><p><i>Loukas Gouskos is a staff researcher at CERN working on the CMS experiment. He received his Ph.D. from&nbsp;the University of Athens. His research interests are focused on the search for physics beyond the standard&nbsp;model and the measurement of the Higgs boson properties. Loukas has a prominent role in searches&nbsp;for&nbsp;supersymmetric partners of the top quark, for&nbsp;Higgs boson decays to a pair of charm quarks, and for Higgs&nbsp;boson pair production in the high-momentum regime. In parallel, Loukas has been deeply involved in the&nbsp;developments in missing transverse momentum reconstruction and in jet tagging particularly with novel&nbsp;machine-learning-based algorithms. Lastly, he is active in the CMS Phase 2 program, focusing on the&nbsp;upgrade of the endcap calorimeters with the High Granularity&nbsp;Calorimeter.</i></p><p><strong>Coffee will be served at 10:30.</strong></p>
id cern-2835621
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2022
record_format invenio
spelling cern-28356212022-11-02T22:03:08Zhttp://cds.cern.ch/record/2835621engGouskos, LoukasJet flavour identification using Graph Neural Networks in CMSJet flavour identification using Graph Neural Networks in CMSEP-IT Data Science Seminars<!--HTML--><p>The identification of the flavour of the parton that initiated the spray of stable particles forming a jet is&nbsp;a&nbsp;fundamental component of&nbsp;standard model measurements and searches for new physics with quarks and gluons in the final state. The first jet identification algorithms were developed using classical methods, however since the&nbsp;beginning of LHC Run 2 the development of&nbsp;significantly more advanced methods, making use of machine learning techniques, led to&nbsp;substantial improvement in&nbsp;performance, and hence in the physics reach.&nbsp;Particularly, Graph Neural Networks (GNNs) are emerging as an extremely powerful class of architectures;&nbsp;GNNs provide a much more natural representation for many&nbsp;tasks in High Energy Physics, including the task&nbsp;of jet identification.&nbsp;In today's seminar we will present a new generation of jet identification algorithms developed by CMS, which&nbsp;exploit&nbsp;customised&nbsp;GNNs tailored to the jet identification task, yielding a drastic improvement in performance&nbsp;compared to&nbsp;previous algorithms.&nbsp;</p><p><i>Loukas Gouskos is a staff researcher at CERN working on the CMS experiment. He received his Ph.D. from&nbsp;the University of Athens. His research interests are focused on the search for physics beyond the standard&nbsp;model and the measurement of the Higgs boson properties. Loukas has a prominent role in searches&nbsp;for&nbsp;supersymmetric partners of the top quark, for&nbsp;Higgs boson decays to a pair of charm quarks, and for Higgs&nbsp;boson pair production in the high-momentum regime. In parallel, Loukas has been deeply involved in the&nbsp;developments in missing transverse momentum reconstruction and in jet tagging particularly with novel&nbsp;machine-learning-based algorithms. Lastly, he is active in the CMS Phase 2 program, focusing on the&nbsp;upgrade of the endcap calorimeters with the High Granularity&nbsp;Calorimeter.</i></p><p><strong>Coffee will be served at 10:30.</strong></p>oai:cds.cern.ch:28356212022
spellingShingle EP-IT Data Science Seminars
Gouskos, Loukas
Jet flavour identification using Graph Neural Networks in CMS
title Jet flavour identification using Graph Neural Networks in CMS
title_full Jet flavour identification using Graph Neural Networks in CMS
title_fullStr Jet flavour identification using Graph Neural Networks in CMS
title_full_unstemmed Jet flavour identification using Graph Neural Networks in CMS
title_short Jet flavour identification using Graph Neural Networks in CMS
title_sort jet flavour identification using graph neural networks in cms
topic EP-IT Data Science Seminars
url http://cds.cern.ch/record/2835621
work_keys_str_mv AT gouskosloukas jetflavouridentificationusinggraphneuralnetworksincms