BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Manuel Szewc (University of Cincinnati)
DTSTART:20230307T193000Z
DTEND:20230307T203000Z
DTSTAMP:20260423T024511Z
UID:nhetc/71
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/nhetc/71/">N
 ull Hypothesis Test for Anomaly Detection</a>\nby Manuel Szewc (University
  of Cincinnati) as part of NHETC Seminar\n\n\nAbstract\nIn this talk we pr
 esent a hypothesis test designed to exclude the background-only hypothesis
  for Anomaly detection searchs. Extending Classification Without Labels\, 
 we show that by testing for statistical independence of the two discrimina
 ting dataset regions\, we are able exclude the background-only hypothesis 
 without relying on fixed anomaly score cuts or extrapolations of backgroun
 d estimates between regions. The method relies on the assumption of condit
 ional independence of anomaly score features and dataset regions\, which c
 an be ensured using existing decorrelation techniques. As a benchmark exam
 ple\, we consider the LHC Olympics dataset where we show that mutual infor
 mation represents a suitable test for statistical independence and our met
 hod exhibits excellent and robust performance at different signal fraction
 s even in presence of realistic feature correlations.\n
LOCATION:https://researchseminars.org/talk/nhetc/71/
END:VEVENT
END:VCALENDAR