一种引入归一化常数的多值因果图推理算法

在将原单值因果图推理算法直接应用于多值因果图推理时,存在不严格满足概率推理中的归一性和互斥性要求的严重问题.该文提出的算法采用以下方法成功解决了这一难题：①假定多值因果图中原因节点对结果节点只贡献概率值,且每个贡献是简单相加的关系.即原因节点对结果节点状态的影响是非直接的,原因节点只影响结果节点各状态的概率分布,结果节点的状态由这个状态概率分布随机决定;②引入归一化常数来保证推理过程中的归一性;③通过推导出多值因果图的一个性质,即可以在推理过程中假定指向同一节点的所有连接事件的各状态之间彼此互斥来保证推理过程的互斥性.从而使得算法在推理过程中同一节点的各状态间完备且互斥,保证了推理的正确性.

A Reasoning Algorithm with Unitizing Coefficient in Multi-value Causality Diagram

There is a critical problem in reasoning process of the Multi-value Causality Diagram(MCD) that it does not meet the expectation of consistency and mutex in probabilistic reasoning process if we adopt the original reasoning algorithm of the Single-value Causality Diagram in reasoning process of the MCD.This paper presents a reasoning algorithm which contains three new concepts,they are(1) assuming that cause nodes do not affect result node directly,they only contribute an intensity to the probability distribution of result node;(2) importing the concept of unitizing coefficient;(3) assuming that they are mutex among all states of all linkage events pointing to a same node.By these ways it ensures that the probability integration of all states of one node equals 1 and these states are mutex each other during reasoning process based on the MCD.