BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Generative Models in Particle Physics
DTSTART;VALUE=DATE-TIME:20221202T164500Z
DTEND;VALUE=DATE-TIME:20221202T170000Z
DTSTAMP;VALUE=DATE-TIME:20260722T170250Z
UID:indico-contribution-3231@cern.ch
DESCRIPTION:Speakers: Sergei Mokhnenko ()\, Fedor Ratnikov (NRU Higher Sch
 ool of Economics)\nAt present\, the primary computational problems in part
 icle physics experiments is the amount\nof computing resources to facilita
 te the slow low level simulation of particles passing through\nthe detecto
 r material.\nA promising way to work around this problem driven by the low
  speed of the full low level Geant4 simulation\nis to use a data-driven su
 rrogate generative models instead. Such models may be trained to directly 
 simulate\na higher level detector responses.\nHowever\, evil is in details
 \, not every surrogate model is equally useful from the physics perspectiv
 e.\nIn this talk we present our experience for developing fast simulation 
 models for different use cases\,\nin different conditions\, and with diffe
 rent requirements\, and demonstrate how strongly those details\ndo affect 
 the final solution.\nFuture prospects of fast simulation approaches which 
 are based on using neural networks are also discussed.\n\nhttps://indico.p
 article.mephi.ru/event/275/contributions/3231/
LOCATION:Hotel Intourist Kolomenskoye 4* Moskvorechye 2
URL:https://indico.particle.mephi.ru/event/275/contributions/3231/
END:VEVENT
END:VCALENDAR