BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Approximate Likelihood
DTSTART;VALUE=DATE-TIME:20151110T094500Z
DTEND;VALUE=DATE-TIME:20151110T103000Z
DTSTAMP;VALUE=DATE-TIME:20191207T222517Z
UID:indico-contribution-939867@indico.cern.ch
DESCRIPTION:Speakers: Kyle Stuart Cranmer (New York University (US))\nMost
physics results at the LHC end in a likelihood ratio test. This includes
discovery and exclusion for searches as well as mass\, cross-section\, and
coupling measurements. The use of Machine Learning (multivariate) algorit
hms in HEP is mainly restricted to searches\, which can be reduced to clas
sification between two fixed distributions: signal vs. background. I will
show how we can extend the use of ML classifiers to distributions paramete
rized by physical quantities like masses and couplings as well as nuisance
parameters associated to systematic uncertainties. This allows for one to
approximate the likelihood ratio while still using a high dimensional fea
ture vector for the data. \n\nBoth the MEM and ABC approaches mentioned ab
ove aim to provide inference on model parameters (like cross-sections\, ma
sses\, couplings\, etc.). ABC is fundamentally tied Bayesian inference and
focuses on the “likelihood free” setting where only a simulator is av
ailable and one cannot directly compute the likelihood for the data. The M
EM approach tries to directly compute the likelihood by approximating the
detector.\nThis approach is similar to ABC in that it provides parameter i
nference in the “likelihood free” setting by using a simulator\, but i
t does not require one to use Bayesian inference and it cleanly separates
issues of statistical calibration from the approximations that are being m
ade. The method is much faster to evaluate than the MEM approach and does
not require a simplified detector description. Furthermore\, it is a gener
alization of the LHC experiments current use of multivariate classifiers f
or searches and integrates well into our existing statistical procedures.\
n\nhttps://indico.cern.ch/event/395374/contributions/939867/
LOCATION:CERN 222/R-001
URL:https://indico.cern.ch/event/395374/contributions/939867/
END:VEVENT
END:VCALENDAR