I'm not sure how to solve this question. I understand that it involves belief propagation but beyond that I'm entirely lost.
Generally speaking, we have G1, one machine that two others depending on; let's contextualize it as a central AC unit. So x1 might be availability of cold air. Whereas G2 and G3 are machines that produce specific goods and X2, X3 are the specific goods therein.
For Q1, if we observe that x2 (some good produced by G2) == 100 units, then what is the probability that G1 == 1 (i.e. cool air is being produced by G1)?
The graph should look like :
-> g3 -> x3
x1 <- g1 -|
-> g2 -> x2
And so I gather that I need to insert factors between adjacent variables, then create "messages" of the like MUf1->x1(x1).
But I don't know how to evaluate it to an exact number.
