BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Exact Bayesian Inference for Big Data: Single- and Multi-Core Appr
 oaches - Murray Pollock (University of Warwick)
DTSTART:20170705T123000Z
DTEND:20170705T131500Z
UID:TALK73157@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:<span>Co-authors: Hongsheng Dai		(Essex)\, Paul Fearnhead		(La
 ncaster)\, Adam Johansen		(Warwick)\, Divakar Kumar		(Warwick)\, Gareth Ro
 berts		(Warwick)        <br></span><span><br>This talk will introduce nove
 l methodologies for exploring posterior distributions by modifying methodo
 logy for exactly (without error) simulating diffusion sample paths. The me
 thodologies discussed have found particular applicability to "Big Data" pr
 oblems. We begin by presenting the Scalable Langevin Exact Algorithm (ScaL
 E) and recent methodological extensions (including Re-ScaLE\, which avoids
  the need for particle approximation in ScaLE)\, which has remarkably good
  scalability properties as the size of the data set increases (it has sub-
 linear cost\, and potentially no cost as a function of data size). ScaLE h
 as particular applicability in the &ldquo\;single-core&rdquo\; big data se
 tting - in which inference is conducted on a single computer. In the secon
 d half of the talk we will present methodology to exactly recombine infere
 nces on separate data sets computed on separate cores - an exact version o
 f &ldquo\;divide and conquer". As such this approach has particu lar appli
 cability in the &ldquo\;multi-core&rdquo\; big data setting. We conclude b
 y commenting on future work on the confluence of these approaches. Joint w
 ork with Hongsheng Dai\, Paul Fearnhead\, Adam Johansen\, Divakar Kumar\, 
 Gareth Roberts.</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR