BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Reinforcement Learning in continuous state-spaces - Philip Sterne 
 (University of Cambridge)
DTSTART:20071126T111500Z
DTEND:20071126T121500Z
UID:TALK9298@talks.cam.ac.uk
CONTACT:Philip Sterne
DESCRIPTION:For this journal club I am interested in continuous state-spac
 e reinforcement learning.  It is my gut feeling that if one makes enough s
 implifying assumptions then the problem becomes solvable (if computational
  concerns are ignored).\n\nIt would be great if everyone attending could s
 pend some time thinking about which assumptions would make the solution ea
 sier (yet still interesting) and what form the various (intractable) integ
 rals would have.\n\nIf you believe that Gaussian Processes might provide a
  good starting point for your thoughts\, have a look at the following pape
 rs (though they are not essential if you have your own ideas)\n\n* "Bayes 
 meets Bellman":http://www.cs.ualberta.ca/~yaki/papers/icml_gptd.ps \n* " R
 einforcement Learning with Gaussian Processes":http://www.cs.ualberta.ca/~
 yaki/papers/gprl_icml05.pdf\n\nThere is also a relevant video lecture by E
 ngel available "here":http://videolectures.net/icml07_engel_gptd/\n
LOCATION:TCM Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR