CRI: Infrastructure for Multi-Agent Decision-Making Research

项目名称：CRI：多智能体决策基础设施研究建议：CNS 0453923PI：Stuart M.ShieberInstitution：哈佛大学项目：该项目将支持社区资源软件的开发和分发，以支持多智能体决策中的研究和教育。PI将扩展一项研究，用于多代理任务的研究、建模和战略开发。试验床是基于一个被称为彩色轨迹(CT)的参数化游戏系列，在这些游戏中，作为个人或团队的代理做出关于资源分配的决定。社区资源软件将提供比以前可用的系统更现实的决策环境，将有助于分析个人、团队和计算机代理的决策策略或这些策略的组合，并使决策策略的开发和测试成为可能。更广泛的影响包括促进多智能体系统和机器学习的研究，支持这些领域的研究和教育的新群体，开发研究和教育材料，支持REU学生，以及参与多样性努力。

AbstractProgram: NSF 04-588 CISE Computing Research InfrastructureTitle: CRI: Infrastructure for Multi-Agent Decision-Making Research Proposal: CNS 0453923PI: Stuart M. ShieberInstitution: Harvard University Project: This project will support the development and distribution of community resource software supporting research and education in multi-agent decision-making. The PI's will expand a research for use in research studies, modeling, and strategy development on multi-agent tasks. The testbed is based on a parameterized family of games called Colored Trails (CT) in which agents as individuals or teams make decisions about resource allocation. The community resource software will provide more realistic decision making settings than previously available systems, will facilitate analysis of decision making strategies of individuals, teams, and computer agents or combinations of these and enable development and testing of decision-making strategies. Broader impacts include facilitating research in multi-agents systems and machine learning, supporting new groups in research and education in these areas, development of research and education materials, support for REU students, and participation in diversity efforts.