BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Sparse trees and long memory: Bayesian inference for discrete time
  series - Ioannis Kontoyiannis\, Athens University of Economics and Busine
 ss
DTSTART:20170419T093000Z
DTEND:20170419T102000Z
UID:TALK71696@talks.cam.ac.uk
CONTACT:Louise Segar
DESCRIPTION:We discuss novel methodological tools for effective Bayesian i
 nference and model selection for general discrete time series data. The st
 arting point of our approach is the use of a rich class of Bayesian hierar
 chical models\, and the observation that the so-called "context tree weigh
 ting" algorithm developed by Willems and co-authors in the early 1990s in 
 the information-theoretic literature\, admits broad extensions that provid
 e effective computational tools for inference in very general settings. We
  will introduce a new class of priors on variable-memory Markov models\, a
 n MCMC Metropolis-within-Gibbs sampler for exploring the full posterior di
 stribution on model space\, and an unusual family of exact algorithms for 
 inference.\n\nApplications range from the classical tasks of estimation an
 d model selection to more application-specific problems including segmenta
 tion\, anomaly detection\, entropy estimation\, causality testing\, and on
 -line prediction with "big data." Our algorithmic\, methodological and the
 oretical results are illustrated by extensive computational experiments on
  both synthetic and real data. Specific applications to data compression\,
  neuroscience\, finance\, genetics\, and animal communication will be ment
 ioned briefly.\n
LOCATION:Engineering Department - Lecture Room - LR6
END:VEVENT
END:VCALENDAR