BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Machine Learning @ CUED
SUMMARY:Information-Theoretic Bounded Rationality - Pedro
A. Ortega (University of Pennsylvania)
DTSTART;TZID=Europe/London:20150907T110000
DTEND;TZID=Europe/London:20150907T120000
UID:TALK60422AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/60422
DESCRIPTION:In this talk I provide an overview of information-
theoretic bounded rationality for planning in sequ
ential decision problems. I show how to ground the
theory on a stochastic computation model for larg
e-scale choice spaces and then derive the free ene
rgy functional as the associated variational princ
iple for characterising bounded-rational decisions
. These decision processes have three important pr
operties: they trade off utility and decision comp
lexity\; they give rise to an equivalence class of
behaviourally indistinguishable decision problems
\; and they possess natural stochastic choice algo
rithms. I will discuss a general class of bounded-
rational sequential planning problems that encompa
sses some well-known classical planning algorithms
as limit cases (such as Expectimax and Minimax)\,
as well as trust- and risk-sensitive planning. Fi
nally\, I will point out formal connections to Bay
esian inference and to regret theory.\n\n*Credits*
: This is joint work with Daniel A. Braun\, Kee-Eu
ng Kim\, Daniel D. Lee\, and Naftali Tishby (in al
phabetical order).
LOCATION:Engineering Department\, CBL Room BE-438
CONTACT:Zoubin Ghahramani
END:VEVENT
END:VCALENDAR