贝叶斯网络中的 OR 关系
如何在贝叶斯网络中表示 OR 关系?例如,P(A | B 或 C)。
我还想知道你如何计算这样的表达式的概率?
先感谢您!
How do you represent an OR-relation in a Bayesian Network? For example, P(A | B OR C).
I also wonder how you can calculate the probability for such an expression?
Thank you in advance!
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
这不是特别适定,因为无法对条件分布中的条件变量求和。然而,一个例子可能会有所帮助。如果我们假设
B
和C
是二元变量并引入变量Z = A或B
。让我们在 P(A,B,C) 上定义以下联合分布现在,根据条件分布的定义,P(A|Z) = P(A,Z)/P (Z)。因此,总结项
和
P(A,Z)
请注意,一旦我们以
Z
为条件,则Z
的两组项保持不变两者之和为 1.0。因此,简而言之,没有计算 P(A|B 或 C) 的通用方法,您需要查看联合分布才能计算适当的概率。
This is not particularly well-posed, because one cannot sum over the conditioned variables in a conditional distribution. However, an example may help. If we assume that
B
andC
are binary variables and introduce a variableZ = A or B
. Let's define the following joint distribution onP(A,B,C)
Now, by the definition of a conditional distribution,
P(A|Z) = P(A,Z)/P(Z)
. So, summing up termsand
P(A,Z)
Notice that once we condition on
Z
that the two sets of terms withZ
held constant both sum to 1.0.So, in short, there isn't a generic way of calculating
P(A|B or C)
, you need to look at the joint distribution in order to calculate the appropriate probabilities.