BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Use of Artificial Neural Network for Event Reconstruction in FARIC
 H detector
DTSTART;VALUE=DATE-TIME:20211116T121000Z
DTEND;VALUE=DATE-TIME:20211116T121500Z
DTSTAMP;VALUE=DATE-TIME:20260613T060219Z
UID:indico-contribution-2320@indico.inp.nsk.su
DESCRIPTION:Speakers: Sergey Kononov (BINP)\nFocusing Aerogel RICH (FARICH
 ) Detector employs a non-uniform aerogel radiator to measure velocity of c
 harged particles with high precision and identify them in the momentum ran
 ge of a few GeV/c. PID system based on FARICH with SiPM readout is propose
 d for the SCTF detector which should provide pi/K separation in the entire
  momentum region of the experiment and mu/pi separation up to 1.5 GeV/c mo
 mentum. \nSiPMs are known to have very high dark count rate of 100 kHz/mm^
 2 at room temperature that could spoil the performance of the FARICH detec
 tor.\nArtificial Neural Networks (ANN) can be used in reconstruction algor
 ithms in High Energy Physics where complex calculations are needed to meas
 ure a signal in presence of background.\nWe employ a fully connected neura
 l network to reconstruct particle's velocity in a FARICH detector. For tra
 ining and testing Geant4 simulated data are used. Velocity resolution of A
 NN was found to be comparable with a geometrical-based reconstruction algo
 rithm in case of no background. ANN provides almost the same resolution fo
 r high momentum particles even in presence of substantial background from 
 SiPM dark counts up to 1 MHz/mm^2.\n\nhttps://indico.inp.nsk.su/event/62/c
 ontributions/2320/
LOCATION:
URL:https://indico.inp.nsk.su/event/62/contributions/2320/
END:VEVENT
END:VCALENDAR