Probability calculation with bayesian belief network

34 Views Asked by At

Trying to calculate the $P(A|\bar{C})$ based on below BBN and wondering if someone can help me confirm the correct setup from below approaches.

enter image description here

Approach 1:

$P(A|\bar{C}) = \sum_{B,D,E} P(B|\bar{C})P(D|\bar{C})P(E|\bar{C})P(A|B)$

I feel like this approach is a non-starter because ($P(B|\bar{C})P(D|\bar{C})P(E|\bar{C})$) implies B,D and E are conditionally independent given C which according to BBN is wrong.

Approach 2:

$P(A|\bar{C}) = \frac{P(\bar{C}|A) P(A)}{P(\bar{C})}$

then $ P(\bar{C}|A) = \sum_{B,D,E} P(\bar{C}|B,D,E)P(A)P(B|A) $

Approach 3:

$P(A|\bar{C}) = \sum_{B,D,E} P(B,D,E | \bar{C})P(A|B) $