基于CBR的贝叶斯网络构造

Bayesian Networks Construction Based on CBR

从数据中学习贝叶斯网络往往会因为搜索空间庞大而耗费大量的时间,所以在构造贝叶斯网络的时候,常依靠以前的经验和知识。该文将过去的贝叶斯网络决策模型保存到案例中,定义相似度和背离度两个衡量指标,在构造新模型时,用基于案例推理的方法检索最为接近的案例,从而进行模型的复用,有效地提高建模的效率。

Learning the structure of Bayesian network from data can be time expensive because of huge search space.When constructing a Bayesian network,people usually depend on the foregone experience and knowledge.This paper stores the Bayesian networks in a case base,and defines two measures:similarity ratio and difference ratio,then uses case-based reasoning method to find the nearest case when modeling a new network.The order is to reuse models so as to improve the efficiency of modeling.