BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Andres Perez (IFLP-Buenos Aires)
DTSTART:20210325T150000Z
DTEND:20210325T160000Z
DTSTAMP:20260423T003257Z
UID:IFTWeb/41
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/IFTWeb/41/">
 Hunting Dark Matter Signals at the LHC with neural networks</a>\nby Andres
  Perez (IFLP-Buenos Aires) as part of IFT-Madrid Webinars\n\n\nAbstract\nW
 e study several simplified dark matter models and their signatures at the\
 nLHC using Neural Networks. We focus on the usual monojet plus missing\ntr
 ansverse energy channel\, but to train the algorithms we organize the data
 \nin 2D histograms instead of event-by-event arrays. This results in a hug
 e\nperformance boost to distinguish between SM only and SM plus new physic
 s\nsignals. We found that Neural Network results do not change with lumino
 sity\,\nif they are shown as a function of S/Sqrt[B]\, where S and B are t
 he number of\nsignal and background events per histogram\, respectively. T
 o keep a broader\napproach\, we do not specify the simplified models coupl
 ing values. This\nprovides flexibility to the method\, since testing a par
 ticular model is\nstraightforward\, only the new physics monojet cross-sec
 tion is needed.\nFurthermore\, we discuss the performance of the networks 
 under wrong\nassumptions. Finally\, we propose multimodel classifiers to s
 earch and\nidentify new signals in a model independent way\, for the next 
 LHC run.\n
LOCATION:https://researchseminars.org/talk/IFTWeb/41/
END:VEVENT
END:VCALENDAR