Description
In the old days, a computer system was adequately modeled as consisting of a computer and a user with one common goal, namely the execution of some specific computation. This is less true today. It can be argued that modern computer and computing systems are better modeled as multi-agent systems consisting of numerous computational agents with partially conflicting goals and interests. That is, they are better modeled as games.As an example, consider the case of sponsored search, where a search engine company sells the ad space of a each page of search results to advertisers. We want to understand which algorithms and protocols are appropriate to use for the search engine company (trying to maximize revenue and consumer satisfaction) and how to write appropriate "bidding agents" for the advertisers (trying to maximize their self-interest). The situation is appropriately modeled as a complex multi-criteria optimization problem with the added difficulty that different agents are trying to optimize different utilities.
Algorithmic Game Theory is an emerging field at the intersection of game theory and computer science aiming at providing the right tools for analyzing these new situations. We present examples of this, focusing at ongoing work at the Center for Algorithmic Game Theory, a new project at the computed science department, funded by the Carlsberg foundation.
| Period | 5 Oct 2007 |
|---|---|
| Event title | Daimi Friday Lectures |
| Event type | Conference |
| Organiser | |
| Location | DenmarkShow on map |