摘要
贝叶斯网络因其处理不确定性问题的能力及其良好的因果推理机制已成为因果数据采掘的主要技术,同样,它也是实现分布估计算法的一个重要途径。而应用贝叶斯网络所解决的一些实际问题的知识表示中会用到混合进制数。因此,给出了十进制数与混合进制数之间的相互转化定理及其严密的理论证明,旨在应用定理中简单的公式来代替以往复杂的转化算法。
Because of the ability to solve indeterminate problems and do well in the causal reasoning, Bayesian network has been the primary technique for causal datum mining, and an important way to realize the estimation of distribution algorithms. When applying the Bayesian network technique on practical problems, the mixed scale numerals are used. Theorem about reciprocal transformation be- tween decimal numerals and mixed scale numerals is discussed in detail and the rigorous proof about this theory is given. The aim is that simple formulas in theorem is substituted for the former complicated transformation algorithms.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第3期504-505,523,共3页
Computer Engineering and Design
基金
湖北省自然科学基金项目(2002AB040)
孝感学院青年基金项目(Z2007026)
关键词
混合进制
十进制
贝叶斯网络
结构学习
参数学习
mixed scale
decimal system
bayesian network
structure learning
parameter learning
