For the Bayesian network problem given in class on Wednesday (in HW 3 as problem 14), Im getting answers that are close to those given in class but they are not quite right. What am I doing wrong?
The prior probabilities given in class for P(alarm), P(smoke) and P(leaving) were rounded to just four significant digits. As an example, if you use .0267 for the prior probability that there is an alarm, you will be having some round off errors that will cause your answers to be slightly off. And the more you repeatedly use these approximations (of only 4 significant digits) the worse the problem can be. Repeat your computation using the exact values.
Related Questions
- For the Bayesian network problem given in class on Wednesday (in HW 3 as problem 14), Im getting answers that are close to those given in class but they are not quite right. What am I doing wrong?
- For the third part of Problem 6, how do I know which edges to include in the Bayesian belief network?
- Does the non-identity problem block a class of arguments against cloning?