BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Identification of low-energy antiprotons on pi-meson background us
 ing machine learning methods in PAMELA experiment
DTSTART;VALUE=DATE-TIME:20161011T111500Z
DTEND;VALUE=DATE-TIME:20161011T113000Z
DTSTAMP;VALUE=DATE-TIME:20260720T121947Z
UID:indico-contribution-598@cern.ch
DESCRIPTION:Speakers: Anton Lukyanov (PhD student)\nOne of main tasks in P
 AMELA experiment is identification of cosmic ray (CR) antiprotons on diffe
 rent\nkinds of background. One example of such background at low energies 
 are pi-meson particles which are\ngenerated in elements of spectrometer co
 nstruction by high-energy CR protons. We propose an approach\nbased on mac
 hine learning methods\, in particular\, Support Vector Machines (SVM). We 
 use two\ndifferent sets of features for classification: track system featu
 res (12 measurements of ionization\nenergy losses along particle track) an
 d calorimeter features (different combinations of energy\nrelease inside c
 alorimeter along reconstructed particle trajectory).\n\nConstructed classi
 fier showed classification accuracy of 96% for antiprotons and 89% for pi-
 mesons\nwhen rigidity $R$ is up to 2 GeV. When $R \\in (2\, 5]$ classifica
 tion accuracy for pi-mesons is 15%.\nFor evaluation of classification accu
 racy we used k-fold cross-validation method with $k=5$.\n\nhttps://indico.
 particle.mephi.ru/event/4/contributions/598/
LOCATION:Milan Hotel Vivaldi-Boccerini
URL:https://indico.particle.mephi.ru/event/4/contributions/598/
END:VEVENT
END:VCALENDAR