BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Accelerated Bayesian inference using deep learning - Adam Moss (No
 ttingham)
DTSTART:20190612T100000Z
DTEND:20190612T103000Z
UID:TALK126370@talks.cam.ac.uk
CONTACT:Will Handley
DESCRIPTION:I introduce a novel Bayesian inference tool that uses a neural
  network to parameterise efficient Markov Chain Monte-Carlo (MCMC) proposa
 ls.  The target distribution is first transformed into a diagonal\, unit v
 ariance Gaussian by a series of non-linear\, invertible\, and non-volume p
 reserving flows. Neural networks are extremely expressive\, and can transf
 orm complex targets to  a simple latent representation from which one can 
 efficiently sample.  Using this method\, I develop a nested MCMC sampler\,
  finding excellent performance on highly curved and multi-modal analytic l
 ikelihoods.  I also demonstrate it on Planck 2015 data\, showing accurate 
 parameter constraints\, and calculate the evidence for simple one-paramete
 r  extensions to LCDM in ~20 dimensional parameter space.
LOCATION:Battcock Tea area
END:VEVENT
END:VCALENDAR