Design of sophisticated Bayesian network systems based on large-scale random fields

在本研究项目中，我们实现了基于大规模随机场的统计推理系统设计的复杂方法的目标。我们所取得的成果的关键是在统计力学中的团簇变分方法中应用了一些技术。当本项目于2006年开始时，它们还不是概率信息处理和统计学习理论研究领域中信念传播的主要技术。我们在研究项目中的突出成果是出版了大规模随机场贝叶斯网络建模和设计的一般方法系统化一书。

In the research project, we have achieved our goal that we construct the sophisticated methodologies for the design of statistical inference systems based on large scale random fields. The key points of our achievements are to employ some technologies in the cluster variation method in the statistical mechanics. They were not so major techniques of belief propagations in the research felids of probabilistic information processing and statistical learning theory when the present project started in 2006. Our outstanding result in our research project is to publish the book that the general methodology for modeling and designing the Bayesian networks with large scale random fields is systematized.