English subtitles

← 32-03 Question 2

32-03 Question 2

Get Embed Code
2 Languages

Showing Revision 3 created 07/12/2017 by Maya Bergamasco.

  1. So here's a Bayes Network,
  2. with 6 variables,
  3. A, B, C, D, E, and F.
  4. And I'd like you to count parameters.
  5. If this was a binary based network,
  6. where each variable
  7. can take on two values
  8. then, A would require
  9. one independent parameters
  10. and B another one.
  11. And C would require
  12. four independent parameters
  13. because there's four different ways
  14. A and B can come together
  15. in condition C.
  16. Now in this question
  17. I'd like to ask you,
  18. what happens if each node
  19. can assume three values,
  20. not just two?
  21. So A can be, A1, A2, A3.
  22. And C can be, C1, C2, C3.
  23. For each node,
  24. specify the number of
  25. independent parameters
  26. required to state
  27. the conditional probability
  28. of that node.
  29. And I'll tell you this is a
  30. tricky question
  31. So for A, the correct answer is two.
  32. I won't give you the other ones.
  33. And, it's two because A
  34. can take three values
  35. but it takes two independent parameters.
  36. The last one can be inferred from
  37. one minus the first two
  38. Please fill in the values
  39. for all the other variables.