BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Generative Networks for Simulation and Statistic for Generative Ne
 tworks
DTSTART;VALUE=DATE-TIME:20260907T113000Z
DTEND;VALUE=DATE-TIME:20260907T121000Z
DTSTAMP;VALUE=DATE-TIME:20260922T191400Z
UID:indico-contribution-2043@events.ncbj.gov.pl
DESCRIPTION:Speakers: Sascha Cassandra Diefenbacher (ITP Heidelberg)\nThe 
 rapidly increasing demands of LHC physics pose a major challenge for Monte
  Carlo simulation. With the High-Luminosity LHC expected to deliver an ord
 er of magnitude more data\, generating sufficiently large and high-fidelit
 y simulated datasets will become increasingly computationally demanding. R
 ecent developments demonstrate that generative machine learning can provid
 e substantial computational speed-ups while retaining the accuracy require
 d for physics analyses.\n\nHowever\, the potential of generative simulatio
 n goes beyond simply replacing an expensive simulation step. Once a genera
 tive model is trained on a finite Monte Carlo dataset\, it can be sampled 
 arbitrarily many times\, raising a fundamental question: how many of these
  generated events are statistically useful? The concept of generative ampl
 ification addresses precisely this question by comparing the statistical p
 ower of generated samples with that of the original training data.\n\nI pr
 esent a framework for quantifying generative amplification without requiri
 ng access to large independent holdout samples. Two complementary approach
 es are introduced: an averaging measure based on phase-space integrals and
  a differential measure based on hypothesis testing. Applied to generative
  models for LHC event generation\, these methods allow us to identify regi
 ons of phase space in which generated samples can provide genuine statisti
 cal amplification\, as well as regions where model uncertainties limit the
  effective sample size.\n\nhttps://events.ncbj.gov.pl/event/468/contributi
 ons/2043/
LOCATION:
URL:https://events.ncbj.gov.pl/event/468/contributions/2043/
END:VEVENT
END:VCALENDAR