Homepage › Solution manuals › Kevin P. Murphy › Machine Learning: a Probabilistic Perspective › Exercise 26.1 - Causal reasoning in the sprinkler network
Exercise 26.1 - Causal reasoning in the sprinkler network
Answers
For question (a) and (b), a perfect intervention cuts out links to , so the probability that the sprinkler is on is:
so is that in case the sprinkler is off.
For question (c), the probability is 0.1, since interfering with the root node of a PGM affectes neither the graphical structure, nor the information propagation within.