Distributed Optimization to assess and bound coordinated actions in Multi-Agent Systems

Supervisor
Meritxell Vinyals - IIIA Spain
Date and time
Tuesday, May 17, 2011 at 4:45 PM - 16:45 rinfresco; ore 17:00 inizio seminario
Place
Ca' Vignal - Piramide, Floor 0, Hall Verde
Programme Director
Alessandro Farinelli
External reference
Publication date
May 10, 2011
Department
Computer Science  

Summary

In this talk I will focus on Distributed Constraint Optimization as an approach to coordinate the actions of a network of cooperative agents.

Recent research in the area of Distributed Constraint Optimization is motivated by some challenging new large-scale applications, such as the control of environmental monitoring sensor networks, the deregulation of power networks or coordination of emergency units in disaster management.

The scale of these new domains requires incomplete algorithms that sacrifice optimality in favour of fast suboptimal solutions.

Because this lack of optimality guarantees is a situation that agents will need to face more and more often in the near future, we are required to design algorithms that allow agents to select actions that result in coordinated behaviour, although maybe not optimality.

In the first part of this talk, I will argue that quality assessment is crucial in this endeavour, from an agent and a system designer perspective, in order to weigh the cost of coordination against its quality.

In the second part of this talk, I will explain which are the insights and techniques that use some of the state-of-the-art algorithms, and frameworks, that exploit the structure of the network of agents to assess and bound coordination in these challenging domains.






© 2002 - 2021  Verona University
Via dell'Artigliere 8, 37129 Verona  |  P. I.V.A. 01541040232  |  C. FISCALE 93009870234