Cargando…

Variational Dropout Sparsification for Particle Identification speed-up

Accurate particle identification (PID) is one of the most important aspects of the LHCb experiment. Modern machine learning techniques such as neural networks (NNs) are efficiently applied to this problem and are integrated into the LHCb software. In this research, we discuss novel applications of n...

Descripción completa

Detalles Bibliográficos
Autores principales: Ryzhikov, Artem, Derkach, Denis, Hushchyn, Mikhail
Lenguaje:eng
Publicado: 2020
Materias:
Acceso en línea:https://dx.doi.org/10.1088/1742-6596/1525/1/012099
http://cds.cern.ch/record/2709467
_version_ 1780965059704389632
author Ryzhikov, Artem
Derkach, Denis
Hushchyn, Mikhail
author_facet Ryzhikov, Artem
Derkach, Denis
Hushchyn, Mikhail
author_sort Ryzhikov, Artem
collection CERN
description Accurate particle identification (PID) is one of the most important aspects of the LHCb experiment. Modern machine learning techniques such as neural networks (NNs) are efficiently applied to this problem and are integrated into the LHCb software. In this research, we discuss novel applications of neural network speed-up techniques to achieve faster PID in LHC upgrade conditions. We show that the best results are obtained using variational dropout sparsification, which provides a prediction (feedforward pass) speed increase of up to a factor of sixteen even when compared to a model with shallow networks.
id cern-2709467
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2020
record_format invenio
spelling cern-27094672022-03-05T04:04:07Zdoi:10.1088/1742-6596/1525/1/012099http://cds.cern.ch/record/2709467engRyzhikov, ArtemDerkach, DenisHushchyn, MikhailVariational Dropout Sparsification for Particle Identification speed-upcs.LGComputing and Computersphysics.data-anOther Fields of PhysicsAccurate particle identification (PID) is one of the most important aspects of the LHCb experiment. Modern machine learning techniques such as neural networks (NNs) are efficiently applied to this problem and are integrated into the LHCb software. In this research, we discuss novel applications of neural network speed-up techniques to achieve faster PID in LHC upgrade conditions. We show that the best results are obtained using variational dropout sparsification, which provides a prediction (feedforward pass) speed increase of up to a factor of sixteen even when compared to a model with shallow networks.arXiv:2001.07493oai:cds.cern.ch:27094672020-01-21
spellingShingle cs.LG
Computing and Computers
physics.data-an
Other Fields of Physics
Ryzhikov, Artem
Derkach, Denis
Hushchyn, Mikhail
Variational Dropout Sparsification for Particle Identification speed-up
title Variational Dropout Sparsification for Particle Identification speed-up
title_full Variational Dropout Sparsification for Particle Identification speed-up
title_fullStr Variational Dropout Sparsification for Particle Identification speed-up
title_full_unstemmed Variational Dropout Sparsification for Particle Identification speed-up
title_short Variational Dropout Sparsification for Particle Identification speed-up
title_sort variational dropout sparsification for particle identification speed-up
topic cs.LG
Computing and Computers
physics.data-an
Other Fields of Physics
url https://dx.doi.org/10.1088/1742-6596/1525/1/012099
http://cds.cern.ch/record/2709467
work_keys_str_mv AT ryzhikovartem variationaldropoutsparsificationforparticleidentificationspeedup
AT derkachdenis variationaldropoutsparsificationforparticleidentificationspeedup
AT hushchynmikhail variationaldropoutsparsificationforparticleidentificationspeedup