BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Partially Observable Markov Decision Processes (POMDPs) - Rowan Mc
 Allister (University of Cambridge)\, Alex Navarro
DTSTART:20160602T133000Z
DTEND:20160602T150000Z
UID:TALK65407@talks.cam.ac.uk
CONTACT:Yingzhen Li
DESCRIPTION:Partially observable Markov decision processes (POMDPs) are a 
 general framework for sequential decision-making tasks when have uncertain
 ty about the state of the world. Many important real-world problems can be
  characterised as POMDPs including financial trading\, spoken dialogue sys
 tems and autonomous car navigation. In recent years reinforcement learning
  has also been characterised as a POMDP opening up an entire literature of
  POMDP solutions to solving the infamous exploration-exploitation dilemma.
  In this talk we begin with simpler models including the Markov process an
 d Hidden Markov models before moving onto their generalisation of POMDPs. 
 We show an important property of POMDP - that they are piecewise-linear an
 d convex - a property which many POMDP algorithms exploit. Alternatively\,
  tree-based search methods can be suitable online-yet-approximate POMDP so
 lutions. Overall we aim to convey some of the main ideas discovered over t
 he past 50 years in POMDP research and to bring the audience up to the pre
 sent day POMDP literature.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR