ART : Agent Reputation Testbed Competition
The essential problem posed by this competition is one that we all face in daily life. That is, how do we make good decisions in the face of uncertainty? Part of the answer to that question lies in making good assumptions. We are use to thinking of assumptions as bad things because we tend only to notice them when they are exposed in a negative light. However, that fact remains that, without taking risks on our assumptions, we would never achieve anything.
With artificial intelligence, there is no exception. Many of the real life problems that we target here at Southampton require computers to make decisions, despite not having all the relevant information available. The type of problem that inspired this competition is just one example of this, in which our agent had to identify which of its competitors provided reliable information, decide how best to use that information, and how best to spend its income to make a profit. With this in mind, our starting point had to be good assumptions about our competitors' strategies, based on careful consideration of the rules of the game. The rest of the work involved churning those assumptions though the mathematical machinery of statistics, and implementing the solutions we got out.
Type: Unofficial Project - Worked on by staff or students in the school but not part of the schools research.
Research Group: Intelligence, Agents, Multimedia Group
Themes: Trust and Provenance, E-Business Technologies
Dates: 1st January 2006 to 1st June 2008
You can edit the record for this project by visiting http://secure.ecs.soton.ac.uk/db/projects/editproj.php?project=363