BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Nornalizing Flows for cosmology applications - Uros Seljak (Berkel
 ey)
DTSTART:20211025T150000Z
DTEND:20211025T160000Z
UID:TALK162241@talks.cam.ac.uk
CONTACT:James Bonifacio
DESCRIPTION:Normalizing Flows (NF) are bijective maps from the data to a G
 aussian (normal) distribution or viceversa. In contrast to other generativ
 e models \nthey are lossless and provide data likelihood via the Jacobian 
 of the transformation. I will first present a novel Sliced Iterative NF (S
 INF)\, \nwhich is based on Optimal Transport theory\, achieving state of t
 he art results in density estimation for small data samples and in anomaly
  detection applications in high energy physics. \nI will discuss its appli
 cations to Bayesian Inference and to Global Optimization problems\, where 
 it enables new methods of sampling and optimization\, which have the poten
 tial to accelerate standard Monte Carlo Markov Chains. In the second half 
 of the talk I will present a Normalizing Flow for data structures with Rot
 ational and Translational Equivariance  (TRENF)\, which can be used for ge
 nerative modeling and likelihood analysis of cosmological data. By trainin
 g the data likelihood on the posterior this approach enables near optimal 
 cosmological likelihood analysis\, where information from all the data is 
 optimally combined into a single number (likelihood) as a function of cosm
 ological parameters. This method provides uncertainty quantification via t
 he full posterior of cosmological parameters\, which paves the way for a c
 omplete and optimal cosmological data analysis with Normalizing Flows.
LOCATION:CMS\, Pav. B\, CTC Common Room (B1.19) [Potter Room]
END:VEVENT
END:VCALENDAR