MIT 6.034 Artificial Intelligence, Fall 2010

View the complete course: http://ocw.mit.edu/6-034F10

Instructor: Patrick Winston

We begin with a review of inference nets, then discuss how to use experimental data to develop a model, which can be used to perform simulations. If we have two competing models, we can use Bayes' rule to determine which is more likely to be accurate.

License: Creative Commons BY-NC-SA

More information at http://ocw.mit.edu/terms

More courses at http://ocw.mit.edu

15:40 These "events" are not generated at random. You have a probability for every variable. P(B) = 0.09, P(R) = 0.5, P(D) = 0.36

Thank you for this lecture!

In the part where you're explaining the bottom-to-top approach starting minute 9:36 (chewing variables from the bottom), I noticed that you omitted the R variable from P(B | T , R) since B has no parents and is therefore not dependent on both T and R. Intuitively, this all makes sense to me, but there is the "explaining away" principle that links B to R since they both cause D. Given that D is correct, there is a relation between B and R. In other words, given that the dog barked, if there is a burglar, this explains away the theory of a Raccoon being present (and vice versa). My question is when and how is this "explaining away" principle used when modeling a system using belief networks? And if we are to use it, how is this relationship between B and R modeled?

I would appreciate any input on this đź™‚

This is beautiful!

Thank you Prof.Winston for this lecture.

starting minute 38 the professor shows that multiplying the probabilities of heads and tails will get us to a point where the probability of the fair coin is bigger (if we get an equal number of heads and tails). However, on each new sample this number will get smaller and smaller since we are multiplying fractions.

however, later on in the demonstration (5 coins) i noticed that the probability of the right coin is getting increased till it hits 1.

I didn't understand this part, if anyone can explain what happened i would appreciate it?

This is gold.

Wait, arent R depended on B based on Markov blanket ?? @minute 9

https://en.wikipedia.org/wiki/Markov_blanket

thank you Prof.Winston