UTCS/AI: Dr. Eyal Amir/University of Illinois Urbana-Champaign Tractable Decision Making in Some Partially Observable Domains ACES 2.402 Friday February 29 2008 11:00 a.m.

Contact Name: 
Jenna Whitney
Date: 
Feb 29, 2008 11:00am - 12:00pm

There is a sign-up schedule for this event
http

://www.cs.utexas.edu/department/webevent/utcs/events/cgi/list_events.cgi

Type of Talk: UTCS Colloquium/AI

Speaker/Affiliation: Dr. Eya

l Amir/University of Illinois Urbana-Champaign

Date/Time: Friday

February 29 2008 11:00 a.m.

Location: ACES 2.402

Host: Vl

adimir Lifschitz

Title: Tractable Decision Making in Some Partially

Observable Domains

Talk Abstract:
Many complex domains offer limi

ted information about their exact state
and the way actions affect them

. There autonomous agents need to
make decisions at the same time that
they learn action models and track
the state of the domain. This combi

ned problem can be represented
within the framework of reinforcement le

arning in POMDPs and is known
to be computationally difficult. In this
presentation I will describe a new
framework for such decision making

learning and tracking. This framework
applies results that we achieved

about updating logical formulas (belief
states) after deterministic act

ions. It includes algorithms that represent
and update the set of possi

ble action models and world states compactly
and tractably. It makes a

decision with this set and updates the set after
taking the chosen act

ion. Most importantly and somewhat surprisingly the
number of actions

that our framework takes to achieve a goal is bounded
polynomially by th

e length of an optimal plan in a fully observable fully
known domain

under lax conditions. Finally our framework leads to a
new stochastic-

filtering approach that has better accuracy than previous
techniques.

Speaker Bio:
Eyal Amir is an Assistant Professor of Computer Scienc

e at the University
of Illinois at Urbana-Champaign (UIUC) since Januar

y 2004. His research
includes reasoning learning and decision making w

ith logical and
probabilistic knowledge dynamic systems and commonsen

se reasoning.
Before UIUC he was a postdoctoral researcher at UC Berkele

y (2001-2003)
with Stuart Russell and did his Ph.D. on logical reasonin

g in AI with John
McCarthy. He received B.Sc. and M.Sc. degrees in mathe

matics and
computer science from Bar-Ilan University Israel in 1992 an

d 1994
respectively. Eyal is a Fellow of the Center for Advanced Studie

s and of
the Beckman Institute at UIUC (2007-2008) was chosen by IEEE

as one
of the 10 to watch in AI (2006) received the NSF CAREER award (

2006)
and awarded the Arthur L. Samuel award for best Computer Science

Ph.D.
thesis (2001-2002) at Stanford University.