0

Had a midterm in my Artificial Intelligence class on MCMC sampling (is it the same as Gibbs sampling?). I was looking over the solution which I found online (in my midterm it was called MCMC liklihood weighting sampler, but in the attached solution its called Gibbs, is that the same?), and I cannot understand how to get the answers shown.

Solution:

enter image description here

I don't understand how to fill in the values (T/F) given the random values from 0.1-1. I do understand how to get the weights on the right (just taking the value from the bottom most table in the chart). But then I don't understand how to get the final value either.

I can't find any instructions online on how to do this, and what it means (at least not in simple terms that don't over complicate it and confuse me), so I would very much appreciate some guidance on how to make sense of these solutions, and how to execute this process on a different set of probabilities.

Thanks!

Jakub Kotowski
  • 7,411
  • 29
  • 38
Doronz
  • 704
  • 8
  • 21
  • Gibbs sampling is a special case of MCMC. – ziggystar Mar 21 '14 at 08:23
  • It's odd because on the two midterm solutions (one that uses Gibbs sampling and the other liklihood weighting) the answers to the question were the same. (Everything else about the question was the same, just the special case was different. – Doronz Mar 21 '14 at 08:25
  • I suggest you read the Wiki pages on Gibbs sampling and on the general MCMC approach to get an idea. Maybe you also need to refresh on likelihood weighting for Bayesian networks. – ziggystar Mar 21 '14 at 08:48

0 Answers0