改进的EM算法及其在防洪决策中应用

在给定贝叶斯网络结构情况下，利用EM算法及改进的EM算法对防洪决策贝叶斯网络进行参数学习，改进的EM算法通过定义祖先集及计算该祖先集中变量的条件概率，降低期望计算的计算量．应用两种算法对防洪决策贝叶斯网络进行了性能比较，结果表明，改进的EM算法用于贝叶斯网络参数学习和决策支持具有较高的计算速度和精确度．

An improved EM algorithm and its application to flood decision-supporting system

The basic principles of Bayesian probability and Bayesian networks are described. The automated creation of Bayesian networks can be separated into two tasks: structure learning, which consists of creating the structure of the Bayesian networks from the collected data; and parameter learning, which consists of calculating the numerical parameters for a given structure. The parameter learning of the flood decision supporting-system is focused on. The EM algorithm and an improved EM algorithm are discussed and applied to the flood decision Bayesian networks to compare their performance. The results indicate that the improved EM algorithm is more precise than traditional EM algorithm. It is shown that the improved EM algorithm can be used in the parameter learning of Bayesian networks and it is also a good way in decision-supporting system.