Type of Talk: UTCS Colloquium/ AI

Date/Time: Friday, Se

ptember 11, 2009/ 11:00 am

Location: ACES 2.302

Host: Peter

Stone

Talk Title: "Teammates in Ad Hoc Teams or What I did on my Sabb

atical"

Talk Abstract: Teams of agents, defined as agents operating

in the same environment with identical utility functions, are typically de

veloped in a planned, coordinated fashion. However, such coordinated deve

lopment is not always possible. Rather, as deployed agents become more com

mon in robotics, e-commerce, and other settings, there are increasing op

portunities for previously unacquainted agents to cooperate in ad hoc team

settings. In such scenarios, it is useful for individual agents to be able
to collaborate with a wide variety of possible teammates under the philoso

phy that not all agents are fully rational. This talk considers an agent th

at is to interact repeatedly with a teammate that will adapt to this intera

ction in a particular suboptimal, but natural way. We formalize this &quo

t;ad hoc team" framework in two ways. First, in a fully cooperativ

e normal form game-theoretic setting, we provide and analyze a fully-imple

mented algorithm for finding optimal action sequences, prove some theoreti

cal results pertaining to the lengths of these action sequences, and provi

de empirical results pertaining to the prevalence of our problem of interes

t in random interaction settings. Second, we consider a cooperative k-arm

ed bandit in which cooperating agents have access to different actions (arm

s). In this setting we prove some theoretical results pertaining to which a

ctions are potentially optimal, provide a fully-implemented algorithm for

finding such optimal actions, and provide empirical results.

Speake

r Bio: Dr. Peter Stone is an Alfred P. Sloan Research Fellow, Guggenheim

Fellow, Fulbright Scholar, and Associate Professor in the Department of C

omputer Sciences at the University of Texas at Austin. He received his Ph.D
in Computer Science in 1998 from Carnegie Mellon University. From 1999 to

2002 he was a Senior Technical Staff Member in the Artificial Intelligence

Principles Research Department at AT&T Labs - Research. Peter''s resea

rch interests include machine learning, multiagent systems, robotics, an

d e-commerce. In 2003, he won a CAREER award from the National Science Fou

ndation for his research on learning agents in dynamic, collaborative, an

d adversarial multiagent environments. In 2004, he was named an ONR Young

Investigator for his research on machine learning on physical robots. In 20

07, he was awarded the prestigious IJCAI 2007 Computers and Thought award

, given once every two years to the top AI researcher under the age of 35.