BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Identifiability conditions for partially-observed Markov chains - 
 Douc\, R (Telecom SudParis)
DTSTART:20140423T145000Z
DTEND:20140423T152500Z
UID:TALK52126@talks.cam.ac.uk
CONTACT:Mustapha Amrani
DESCRIPTION:Co-authors: Franois ROUEFF (LTCI\, UMR 5141\, Telecom Paristec
 h\, France)\, Tepmony SIM (LTCI\, UMR 5141\, Telecom Paristech\, France) \
 n\nThis paper deals with a parametrized family of partially-observed bivar
 iate Markov chains. We establish that the limit of the normalized log-like
 lihood is maximized when the parameter belongs to the equivalence class of
  the true parameter\, which is a key feature for obtaining consistency of 
 the Maximum Likelihood Estimators in well-specified models. A novel aspect
  of this work is that geometric ergodicity of the Markov chain associated 
 to the complete data\, or exponential separation on measures are no more n
 eeded provided that the invariant distribution is assumed to be unique\, r
 egardless its rate of convergence to the equilibrium. The result is presen
 ted in a general framework including both fully dominated or partially dom
 inated models as Hidden Markov models or Observation-Driven times series o
 f counts.\n
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR