2005 | OriginalPaper | Buchkapitel
Prediction Games
(Extended Abstract)
verfasst von : Amir Ronen, Lyron Wahrmann
Erschienen in: Internet and Network Economics
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
We introduce a new class of mechanism design problems called
prediction games
. There are
n
self interested agents. Each agent
i
has a
private
input
x
i
and a cost of accessing it. Given are a function
f
(
x
1
, ...,
x
n
) which predicts whether a certain event will occur and an independent distribution on the agents’ inputs. Agents can be rewarded in order to motivate them to access their inputs. The goal is to design a mechanism (protocol) which in equilibrium predicts
f
(.) and pays in expectation as little as possible.
We investigate both, exact and approximate versions of the problem and provide several upper and lower bounds. In particular, we construct an optimal mechanism for every anonymous function and show that any function can be approximated with a relatively small expected payment.