BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//6.4.7.2//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:1-359@lptms.universite-paris-saclay.fr
DTSTART:20150504T143000Z
DTEND:20150504T160000Z
DTSTAMP:20150428T071448Z
URL:http://www.lptms.universite-paris-saclay.fr/seminars/tri-seminaire-de-
 physique-statistique-balazs-kegel/
SUMMARY:Tri-Séminaire de Physique Statistique : Balázs Kégel - IPhT\, CE
 A-Saclay\, salle Itzykson - 4 Mai 15 14:30
DESCRIPTION:Learning to discover: signal/background separation and the Higg
 s boson challenge\nBalázs Kégel (Laboratoire de l'Accélérateur Linéai
 re\, Univ. d'Orsay)\nClassification algorithms have been routinely used si
 nce the 90s in high-energy physics to separate signal and background in pa
 rticle detectors. The goal of the classifier is to maximize the sensitivit
 y of a counting test in a selection region. It is similar in spirit but fo
 rmally different from the classical objectives of minimizing misclassifica
 tion error or maximizing AUC. We start the talk by motivating the problem 
 on an ongoing example of detecting the Higgs boson in the tau-tau decay ch
 annel in the ATLAS detector of the LHC. We formalize the problem\, then go
  on by describing the usual analysis chain\, and explain some of the choic
 es physicists make when designing a classifier for optimizing the discover
 y significance. We derive different surrogates that capture this goal and 
 show some simple techniques to optimize them\, raising some questions both
  on the statistical and on the algorithmic side. We end the talk by presen
 ting a data challenge we organized to draw the attention of the machine le
 arning and statistics communities to this important application and to imp
 rove the techniques used to optimize the discovery significance. With a Ph
 D in computer science\, Balázs Kégl has been a researcher in the Linear 
 Accelerator Laboratory of the CNRS and the chair of the Center for Data Sc
 ience of the Université Paris-Saclay since 2014. He has published more th
 an hundred papers on unsupervised and supervised learning\, large-scale Ba
 yesian inference and optimization\, and on various applications. At his cu
 rrent position he has been the head of the AppStat team working on machine
  learning and statistical inference problems motivated by applications in 
 high-energy particle and astroparticle physics.
CATEGORIES:seminars
LOCATION:IPhT\, CEA-Saclay\, salle Itzykson\, CEA - Saclay\, Saclay\, Franc
 e
GEO:48.73668;2.180033999999978
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=CEA - Saclay\, Saclay\, Fra
 nce;X-APPLE-RADIUS=100;X-TITLE=IPhT\, CEA-Saclay\, salle Itzykson:geo:48.7
 3668,2.180033999999978
END:VEVENT
END:VCALENDAR