Very much agree with this post. In my opinion, ch. 14 (Bayes nets) was the most important chapter in AI: AMA; I actually did every single non-programming exercise for that chapter, coming back later to ensure I could redo those I got wrong. I’d “learned” about probability by reading, but it wasn’t until I put my nose to the grindstone and did the math that I started being able to see probability flowing through the networks, that I got S1 intuitions for the difference between independence and conditional independence.
Of course, that was just a chapter—hardly “careful study”; I’m very much looking forward to Pearl.
Very much agree with this post. In my opinion, ch. 14 (Bayes nets) was the most important chapter in AI: AMA; I actually did every single non-programming exercise for that chapter, coming back later to ensure I could redo those I got wrong. I’d “learned” about probability by reading, but it wasn’t until I put my nose to the grindstone and did the math that I started being able to see probability flowing through the networks, that I got S1 intuitions for the difference between independence and conditional independence.
Of course, that was just a chapter—hardly “careful study”; I’m very much looking forward to Pearl.