New Machine Learning Developments in ROOT/TMVA
Document identifier: oai:DiVA.org:ltu-76550
Access full text here:
10.1051/epjconf/201921406014Keyword: Natural Sciences,
Computer and Information Sciences,
Software Engineering,
Naturvetenskap,
Data- och informationsvetenskap,
Programvaruteknik,
Computer Sciences,
Datavetenskap (datalogi),
ROOT,
TMVA,
Machine Learning,
MaskininlärningPublication year: 2019Abstract: The Toolkit for Multivariate Analysis, TMVA, the machine learning package integrated into the ROOT data analysis framework, has recently seen improvements to its deep learning module, parallelisation of multivariate methods and cross validation. Performance benchmarks on datasets from high-energy physics are presented with a particular focus on the new deep learning module which contains robust fully-connected, convolutional and recurrent deep neural networks implemented on CPU and GPU architectures. Both dense and convo-lutional layers are shown to be competitive on small-scale networks suitable for high-level physics analyses in both training and in single-event evaluation. Par-allelisation efforts show an asymptotical 3-fold reduction in boosted decision tree training time while the cross validation implementation shows significant speed up with parallel fold evaluation.
Authors
Kim Albertsson
Luleå tekniska universitet; EISLAB; CERN
Other publications
>>
Sergei Gleyze
University of Florida
Other publications
>>
Marc Huwiler
EPFL
Other publications
>>
Vladimir Ilievski
EPFL
Other publications
>>
Lorenzo Moneta
CERN
Other publications
>>
Saurav Shekar
ETH Zurich
Other publications
>>
Victor Estrade
CERN
Other publications
>>
Akshay Vashistha
CERN. Karlsruhe Institute of Technology
Other publications
>>
Stefan Wunsch
CERN. Karlsruhe Institute of Technology
Other publications
>>
Omar Andres Zapata Mesa
University of Antioquia. Metropolitan Institute of Technology
Other publications
>>
Record metadata
Click to view metadata
header:
identifier: oai:DiVA.org:ltu-76550
datestamp: 2021-04-19T12:55:20Z
setSpec: SwePub-ltu
metadata:
mods:
@attributes:
version: 3.7
recordInfo:
recordContentSource: ltu
recordCreationDate: 2019-10-29
identifier:
http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-76550
10.1051/epjconf/201921406014
titleInfo:
@attributes:
lang: eng
title: New Machine Learning Developments in ROOT/TMVA
abstract: The Toolkit for Multivariate Analysis TMVA the machine learning package integrated into the ROOT data analysis framework has recently seen improvements to its deep learning module parallelisation of multivariate methods and cross validation. Performance benchmarks on datasets from high-energy physics are presented with a particular focus on the new deep learning module which contains robust fully-connected convolutional and recurrent deep neural networks implemented on CPU and GPU architectures. Both dense and convo-lutional layers are shown to be competitive on small-scale networks suitable for high-level physics analyses in both training and in single-event evaluation. Par-allelisation efforts show an asymptotical 3-fold reduction in boosted decision tree training time while the cross validation implementation shows significant speed up with parallel fold evaluation.
subject:
@attributes:
lang: eng
authority: uka.se
topic:
Natural Sciences
Computer and Information Sciences
Software Engineering
@attributes:
lang: swe
authority: uka.se
topic:
Naturvetenskap
Data- och informationsvetenskap
Programvaruteknik
@attributes:
lang: eng
authority: uka.se
topic:
Natural Sciences
Computer and Information Sciences
Computer Sciences
@attributes:
lang: swe
authority: uka.se
topic:
Naturvetenskap
Data- och informationsvetenskap
Datavetenskap (datalogi)
@attributes:
lang: eng
topic: ROOT
@attributes:
lang: eng
topic: TMVA
@attributes:
lang: eng
topic: Machine Learning
@attributes:
lang: swe
authority: ltu
topic: Maskininlärning
genre: Research subject
@attributes:
lang: eng
authority: ltu
topic: Machine Learning
genre: Research subject
language:
languageTerm: eng
genre:
conference/other
ref
note:
Published
10
name:
@attributes:
type: personal
authority: ltu
namePart:
Albertsson
Kim
role:
roleTerm: aut
affiliation:
Luleå tekniska universitet
EISLAB
CERN
nameIdentifier:
ashlaban
0000-0002-5052-9629
@attributes:
type: personal
namePart:
Gleyze
Sergei
role:
roleTerm: aut
affiliation: University of Florida
@attributes:
type: personal
namePart:
Huwiler
Marc
role:
roleTerm: aut
affiliation: EPFL
@attributes:
type: personal
namePart:
Ilievski
Vladimir
role:
roleTerm: aut
affiliation: EPFL
@attributes:
type: personal
namePart:
Moneta
Lorenzo
role:
roleTerm: aut
affiliation: CERN
@attributes:
type: personal
namePart:
Shekar
Saurav
role:
roleTerm: aut
affiliation: ETH Zurich
@attributes:
type: personal
namePart:
Estrade
Victor
role:
roleTerm: aut
affiliation: CERN
@attributes:
type: personal
namePart:
Vashistha
Akshay
role:
roleTerm: aut
affiliation: CERN. Karlsruhe Institute of Technology
@attributes:
type: personal
namePart:
Wunsch
Stefan
role:
roleTerm: aut
affiliation: CERN. Karlsruhe Institute of Technology
@attributes:
type: personal
namePart:
Mesa
Omar Andres Zapata
role:
roleTerm: aut
affiliation: University of Antioquia. Metropolitan Institute of Technology
originInfo:
dateIssued: 2019
publisher: EDP Sciences
relatedItem:
@attributes:
type: host
titleInfo:
title: 23rd International Conference on Computing in High Energy and Nuclear Physics (CHEP 2018)
@attributes:
type: series
titleInfo:
title: EPJ Web of Conferences
identifier: 2100-014X
physicalDescription:
form: print
typeOfResource: text