Renewal of intelligent decision support systems laboratory

由于当今软件的规模和复杂性、它们的关键作用以及所需的可靠性，以及由于计算机芯片和网络的技术进步，通过一组称为代理的协作计算机在大的域中解决复杂问题已经变得必要和可能。根据应用，代理需要基于不确定的局部知识和观察来确定其域的状态(分布式概率推理)，或者识别是否存在满足其局部约束的全局解，如果存在，则找到这样的解(分布式约束满足)，或者决定在面对不确定性(分布式决策)时，什么联合行动能使他们的环境达到最理想的状态。我在9年前创立的智能决策支持系统实验室研究如何建立能够进行分布式概率推理、分布式约束满足和分布式决策的多智能体系统。我们关注的多智能体知识表示是多智能体图形模型，我们强调的性能问题是理论可靠性和计算效率。我们设计和分析了执行这些任务的多智能体算法。并实施和测试原型(目前分发给30个国家的注册用户)。除了获得代理执行任务所需的图形模型外，还在实验室研究从数据中发现知识和从专家那里启发。此外，该实验室上一次大规模采购计算设备是在2001年。由于芯片技术的进步，现有设备的性能现在已经落后。*所要求的设备将用最先进的计算设备更新实验室，以支持新算法的设计和评估以及原型系统的开发。

Due to scales and complexities of today's software, their critical roles and hence reliabilities required, and due to technological advances in computer chips and networking, it has become both necessary and possible to solve a complex problem in a large domain by a set of cooperating computers, called agents.  Depending on applications, agents need to determine what is the state of their domain based on uncertain local knowledge and observations (distributed probabilistic reasoning), or identify whether a global solution exists which satisfies their local constraints and if so find such a solution (distributed constraint satisfaction), or decide what is the joint action that would bring their environment to a most desirable state in the face of uncertainty (distributed decision making).      The Intelligent Decision Support Systems Laboratory that I founded 9 years ago studies how to build multiagent systems that are capable of distributed probabilistic reasoning, distributed constraint satisfaction, and distributed decision making.  Agent knowledge representation that we focus on is multiagent graphical models, and performance issues that we emphasize are theoretical soundness and computational efficiency.  Multiagent algorithms for performing these tasks are designed and analyzed, and prototypes (currently distributed to registered users in 30 countries) are implemented and tested.   To acquire graphical models needed by agents to perform their tasks, knowledge discovery from data and elicitation from expert are also investigated in the lab.    The last major purchase of computing equipment in the lab was in 2001. Due to advancement in chip technology, performance of the existing equipment is now falling behind.  The requested equipment will renew the lab with the state of the art computing equipment to support design and evaluation of new algorithms and development of prototype systems.