BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Unbinned inclusive cross-section measurements with machine-learned
  systematic uncertainties
DTSTART:20260507T120000Z
DTEND:20260507T130000Z
DTSTAMP:20260506T054000Z
UID:indico-event-17874@indico.global
DESCRIPTION:Speakers: Claudius Krause\n\nWe introduce a novel methodology 
 for addressing systematic uncertainties in unbinned inclusive cross-sectio
 n measurements and related collider-based inference problems. Our approach
  incorporates known analytic dependencies on parameters of interest\, incl
 uding signal strengths and nuisance parameters. When these dependencies ar
 e unknown\, as is frequently the case for systematic uncertainties\, dedic
 ated neural network parametrizations provide an approximation that is trai
 ned on simulated data. The resulting machine-learned surrogate captures th
 e complete parameter dependence of the likelihood ratio\, providing a near
 -optimal test statistic. As a case study\, we perform a first-principles i
 nclusive cross-section measurement of $H \\to \\tau\\tau$ in the single-le
 pton channel\, utilizing simulated data from the FAIR Universe Higgs Uncer
 tainty Challenge. Results in Asimov data\, from large-scale toy studies\, 
 and using the Fisher information demonstrate significant improvements over
  traditional binned methods.  Our submission won first place ex aequo in 
 the FAIR Universe Higgs Uncertainty Challenge and is available at https://
 arxiv.org/abs/2505.05544.If time permits\, I will also show how these tech
 niques can be applied to extract the gluon PDF from dileptonic $\\bar t t$
  production at the LHC.\n\nhttps://indico.global/event/17874/
LOCATION:Aula Teorici (University of Milan Statale)
URL:https://indico.global/event/17874/
END:VEVENT
END:VCALENDAR