This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ode model with just the one extra node B. Draw the two possible models they should consider.
Draw only the new node B and its edges — you do not need conditional probability tables (yet):
Alternate Model A Z Z R R T 29 Alternate Model B T Despite two positive test results each, Ulric, Victor, and Wendy all turn out to be healthy. An expedition to (apparently zombie-free) Brainton revealed that 80% of its
population test positive using your town's test. One of the alternate models is very
likely! G (6 points) Assume that you think a model whose Bayes Net representation has k+1 parameters is about half as likely a priori as a model with k parameters, unless k = 0, in which case assume that the model probability is some constant, c. How likely is the original (three variable) model, a priori? How likely are each of the models you drew, a priori?
Alternate Model A Alternate Model B H (7 points)
To rule out your original model entirely, you decide that an alternate model has to explain the data not
View Full Document
This document was uploaded on 03/17/2014 for the course EECS 6.034 at MIT.
- Fall '10
- Artificial Intelligence