Hélène Bestougeff

Modeling Task and Teams through Game Theoretical Agents

Organizations are represented as conversational networks of agents, which interactions are modeled by a two-player game of deterrence, i.e. a qualitative game based on the concept of threshold. A congestion control algorithm is then derived from the game Boolean solution set. Last, analysis is extended to the case where the solution set is fuzzyfied. On the whole, the approach enables a mix-management : decentralized when agents can manage their relations successfully by themselves; centralized when the risk of congestion arises.

Multi-agent systems reliability, fuzziness, and deterrence

The problem of MAS reliability is approached through representing the functioning of a MAS as a system of logical implications, and then interpreting this system as a game of deterrence. The game solutions provide indicators for the agents reliability, and enable in case of an agents failure, to select a search direction for determining the origin of the failure. The MAS reliability is increased by duplicating some agents. The impact of the duplicates positioning in the MAS is analysed on a particular case.

Bridging Games and Diplomacy

The gap between models and reality is an issue covering almost every field of human knowledge, from physics to linguistics, and from economics to biology. On the one hand, quantitative people are mainly concerned with developing tools and consistent theories, even at the price of neglecting aspects of reality that do not fit in the frame or would lead to technical intricacies that might impede well-stated conclusions. On the other hand, observers and practitioners immersed in the “real world,” tend to ignore quantitative tools and theories for reasons ranging from lack of understanding to hostility.

Modeling Traffic Control through Deterrent Agents

Multi Agents Systems (MAS) technology is used to build Urban Traffic Control (UTC) systems based on traffic lights regulation. Congestion avoidance is analyzed and simulated through modeling the network as a game of deterrence, as previously done in the analysis of communication and workflow networks.

Deterrence and Defeasibility in Argumentation Process for ALIS Project

Argumentation issues, which are of core importance to ALIS, are addressed through a particular category of qualitative games called Games of Deterrence. The graphs associated with those games are interpreted as sets of inferences sequences between statements in the framework of non-monotonic logic. Thus an argumentation process is interpreted as a game of deterrence, which resolution determines the truth or falsity of statements, and the possible argumentation strategies of the parties.

Bayesian Networks and Games of Deterrence

The present paper analyzes possible bridges between two types of graphical models: on the one hand, Bayesian Networks, and on the other hand, Graphs of Deterrence associated with a particular category of qualitative games called matrix Games of Deterrence, in which players do not look for optimal outcomes but for acceptable ones. Three related-types of relations are scrutinized: implications and rebuttals; priors and hidden parts of the graph; probability and playability.