BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A new mathematical framework for optimal choice of actions - Emo T
 odorov\, UCSD
DTSTART:20070912T150000Z
DTEND:20070912T160000Z
UID:TALK8026@talks.cam.ac.uk
CONTACT:dw304
DESCRIPTION:Optimal choice of actions is relevant to fields as diverse as 
 neuroscience\, psychology\, economics\, operations research\, computer sci
 ence\, robotics and automation\, aerospace engineering. Despite this broad
  relevance the abstract setting is similar: we have an agent choosing acti
 ons over time\, a dynamical system whose state is affected by those action
 s\, and a cumulative performance criterion which the agent seeks to optimi
 ze. Designing synthetic agents that can solve problems of this kind\, or u
 nderstanding how natural agents manage to do so\, remains hard. The diffic
 ulties are partly due to overly generic problem formulations. Here we prop
 ose a more structured formulation in which optimal choice of actions is gr
 eatly simplified: it is reduced to a linear problem\, in both discrete and
  continuous domains. This facilitates the computation of solutions\, gives
  rise to unique theoretical properties\, and yields original algorithms th
 at solve existing problems faster than Dynamic Programming and Reinforceme
 nt Learning. Discovery of a general class of easily solvable problems tend
 s to motivate researchers to reformulate or approximate their problems wit
 hin the new class. Our results suggest that in many cases this will be pos
 sible. The new framework is likely to find applications in diverse fields 
 of science and engineering.\n\n
LOCATION:LR5\, Department of Engineering\, Trumpington Street
END:VEVENT
END:VCALENDAR