BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Optimal Payments in Dominant-Strategy Mechanisms - Victor Naradits
 kiy\, University of Southampton
DTSTART:20130314T100000Z
DTEND:20130314T110000Z
UID:TALK43889@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Mechanism design is concerned with decision making involving m
 ultiple self-interested participants. Each of the participants holds priva
 te information that is relevant to the quality of a decision. A mechanism 
 designer has an objective such as maximising the total utility of the part
 icipants (utilitarian) or guaranteeing each participant a certain level of
  utility (egalitarian). To make the right decision according to the object
 ive\, the designer needs to reveal private information of the participants
 . Truthful revelation is possible with the help of Groves mechanisms that 
 use monetary payments to ensure the participants cannot benefit by misrepo
 rting their private information. Until recently\, virtually all the attent
 ion had been paid to one Groves mechanism called the Vickrey-Clarke-Groves
  (VCG) mechanism. In the last decade a series of papers started investigat
 ing whether there are other Groves mechanisms that are better than VCG for
  a given scenario and objective.\n\nIn this talk I will present general ch
 aracterisation results along with an approach for choosing the best Groves
  mechanism for a wide class of scenarios and objectives (including utilita
 rian and egalitarian). The characterisation links optimality of mechanism'
 s payments to a geometric condition involving triangulations of hypercubes
 . When this condition is satisfied\, I constructively show existence of an
  optimal payment function that is piecewise linear. I then use the techniq
 ue to derive optimal mechanisms for several fundamental scenarios. The tec
 hnique also provides a unified way to prove a number of results that were 
 previously obtained using different\, more complicated techniques.
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR