BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Belief and Truth in Hypothesised Behaviours - Stefano V. Albrecht 
 -  School of Informatics at The University of Edinburgh
DTSTART:20150910T100000Z
DTEND:20150910T110000Z
UID:TALK60639@talks.cam.ac.uk
CONTACT:Alexandre Khae Wu Navarro
DESCRIPTION:There is a long history in game theory on the topic of Bayesia
 n or "rational" learning\, in which players maintain beliefs over a set of
  alternative behaviours\, or types. This idea has gained increasing intere
 st in the AI community\, where it is used to control a single agent in a s
 ystem composed of multiple agents with unknown behaviours. The idea is to 
 hypothesise a set of types\, each specifying a possible behaviour for the 
 other agents\, and to plan our own actions with respect to those types whi
 ch we believe are most likely\, based on the observed actions. The game th
 eory literature studies this idea primarily in the context of equilibrium 
 attainment. In contrast\, many AI applications have a focus on task comple
 tion and payoff maximisation\, which renders the game theory literature on
  this subject of limited applicability. With this perspective in mind\, we
  identify and address a spectrum of questions pertaining to belief and tru
 th in hypothesised types. We formulate three basic ways to incorporate evi
 dence into posterior beliefs and show when the resulting beliefs are corre
 ct\, and when they may fail to be correct. Moreover\, we demonstrate that 
 prior beliefs can have a significant impact on our ability to maximise pay
 offs in the long-term\, and that they can be computed automatically with c
 onsistent performance effects. Furthermore\, we analyse the conditions und
 er which we are able complete our task optimally\, despite inaccuracies in
  the hypothesised types. Finally\, we show how the correctness of hypothes
 ised types can be ascertained during the interaction via an automated stat
 istical analysis.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR