【摘要】 文章扩展经典的先验BN模型,采用两层学习结构讨论分组样本下BN模型的条件概率及学习算法:一层是对各组私有条件概率分布的学习;另一层是对各组公有条件概率分布的学习。算法在综合公有后验条件概率分布和本组学习实例数据分布特征的基础上,实现对各组私有条件概率分布的学习,并可通过经验或学习来改变综合值中共性和个性的比例。更多还原

【Abstract】 This paper has extended classical prior BN model and discussed BN conditional probabilities &its learning algorithm for group samples with two level learning structure:one level expresses learning private conditional probabilities for each group;another level expresses learning public conditional probability. On the basis of synthesizing public posterior conditional probability distribution and the group learning instance data,the learning private condition probability distribution can be implemented. The proportion of the specific character to the general can be changed by experience or learning.更多还原