Modeling Cooperative Multiagent Problem Solving as Decentralized Decision Processes
نویسندگان
چکیده
We present a formal framework for modeling and control of cooperative multiagent problem solving. Such a framework is crucial for understanding and evaluating various design choices based on first principles. This framework allows us to define precisely the notion of optimal cooperation and analyze its complexity. Communication decisions are introduced to allow agents to reason explicitly about coordination, capturing effectively the decentralized nature of the problem solving process and uncertainty in the system. Because finding optimal control is intractable, we develop several approximations and evaluate their effectiveness. The result is an integrated decision-theoretic framework for both agent planning and
منابع مشابه
Model and analysis of local decision based on cooperative self-organization for problem solving
This paper presents an approach based on cooperative self-organization for artificial systems, in order to tackle openness and dynamics. In this work, cooperation is used as a local criterion enabling parts of the system –the cooperative agents– to reorganize as to autonomously modify their interactions and then the global function. The difficulty in defining cooperation and the means to reesta...
متن کاملAn Instrumentalized Participatory Approach for Cooperative Knowledge Acquisition to Build a Social Mabs
This paper proposes a participatory and cooperative approach to acquire necessary knowledge to build a MultiAgent-Based Simulation (MABS). This approach is based on role-playing, meeting and Computer Supported Cooperative Work (CSCW) principles. During a meeting firm actors simulate problem solving processes by playing their usual roles. The result of several meetings is a corpus which is analy...
متن کاملLearning for Multiagent Decentralized Control in Large Partially Observable Stochastic Environments
This paper presents a probabilistic framework for learning decentralized control policies for cooperative multiagent systems operating in a large partially observable stochastic environment based on batch data (trajectories). In decentralized domains, because of communication limitations, the agents cannot share their entire belief states, so execution must proceed based on local information. D...
متن کاملExploiting Separability in Multiagent Planning with Continuous-State MDPs (Extended Abstract)
Decentralized partially observable Markov decision processes (Dec-POMDPs) provide a general model for decision-making under uncertainty in cooperative decentralized settings, but are difficult to solve optimally (NEXP-Complete). As a new way of solving these problems, we recently introduced a method for transforming a Dec-POMDP into a continuous-state deterministic MDP with a piecewise-linear a...
متن کاملAgent Influence as a Predictor of Difficulty for Decentralized Problem-Solving
We study the effect of problem structure on the practical performance of optimal dynamic programming for decentralized decision problems. It is shown that restricting agent influence over problem dynamics can make the problem easier to solve. Experimental results establish that agent influence correlates with problem difficulty: as the gap between the influence of different agents grows, proble...
متن کامل