Invited Lecture on "Coordinating Multi-agent Systems using Social Laws" by Thomas Ågotnes (U Bergen)

Actividad válida para reconocimiento de créditos.

Se reconocerán 0.2 créditos a los alumnos que asistirán a la conferencia y que redactarán un resumen de la misma (máximo 2 páginas). Para la entrega de los resúmenes, la obtención del certificado de asistencia o cualquier otra información, contactar con marin.lujak@urjc.es Conferencia dentro del marco del Proyecto "Tecnologías del Acuerdo" (Consolider CSD2007-00022).

 
 

Date: September 26th 2013

 
Location: URJC, Campus de Móstoles, Departamental II, Salón de grados
 

Programme:

11:00 - 13:00: Invited Lecture on " Coordinating Multi-agent Systems using Social Laws " by Thomas Ågotnes [pdf]

Abstract:

Social laws (or normative systems) have emerged as a natural and powerful paradigm for coordinating such systems, exposing the whole spectrum between fully centralised and fully decentralised coordination mechanisms. A social law is, intuitively, a constraint on the behaviour of agents, which ensures that their individual behaviours are compatible. In a standard multi-agent state transition diagram (where transitions are labelled with the name of the agent effecting the transition), a social law is simply a labelling of the transitions saying whether or not they are "legal", or "desirable". An illegal transition could, for example, correspond to a component malfunctioning. Different normative systems give rise to different global system properties, assuming that the social law is complied with. Such properties can be specified and analysed in temporal modal logic. However, it might be that not all agents/components comply, either deliberately or because of a malfunction. In the talk I will, in addition to introducing and motivating the idea of social laws, discuss (non-)compliance from two angles. First: when is it rational for an agent to comply? Since the properties of the resulting system depend on compliance of all the agents in the system (think of the norm "drive on the right side of the road"), this requires a game theoretic analysis. Second, how can we identify the most important agents/components in the system, the agents whose compliance is crucial for the proper functioning of the system? I will talk about some resulting decision problems, combining logic, game theory, voting theory and complexity theory. 

 

Speaker: Thomas Ågotnes

 

Thomas Ågotnes is a full Professor of Information Science at the University of Bergen. His main interests is multi-agent systems, in particular the use of formal logic to model, analyze and reason about interaction in multi-agent systems, and he has worked and published extensively in these areas. He was a a co-winner of the best paper award at AAMAS on logical analysis of social laws in 2009. Ågotnes is an active member of the international multi-agent systems community. He is a member of the board of directors of the European Association for Multi-Agent Systems (EURAMAS), a member of the steering committe of the CLIMA workshop series, is a PC chair of LOFT 2014, and was a PC chair of DEON 2012 and the general chair of STAIRS 2010.