# ps2_sol - 1 Introduction to Probabilistic Graphical Models...

This preview shows pages 1–3. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 1 Introduction to Probabilistic Graphical Models Lecturer: Eran Segal Problem set 2 - Solutions 1. General notes: • Please note that ∑ y P ( X | Y = y ) 6 = P ( X ). Moreover, it can be easily verified that ∑ y P ( X | Y = y ) might not even represent a valid distribution. • Conditional probability P ( X,Y | Z ) is not defined for the case P ( Z ) = 0, but the conditional independence ( X ⊥ Y | Z ) is defined to hold for P ( Z ) = 0. In questions (a) and (b) one need to justify divisions by, or conditioning on, such zero probability cases (which in those cases is pretty much straight forward from definitions). In question (c), one utilizes the positivity of the distribution to justify such manipulations. • Recall that ( X ⊥ Y | Z ) holds if and only if P ( X,Y | Z ) = P ( X | Z ) P ( Y | Z ), but also if and only if P ( X | Z ) = P ( X | Y,Z ) (by definition). (a) Answer: Assume: ( X ⊥ Y,W | Z ) (1) = ⇒ P ( X,Y,W | Z ) = P ( X | Z ) P ( Y,W | Z ) Case analysis: i. Case: P ( Z,W ) = 0 = ⇒ trivial from definition of conditional independence. ii. Case: P ( Z ) = 0 = ⇒ P ( Z,W ) = 0 = ⇒ by case i . iii. Case: P ( Z,W ) ,P ( Z ) 6 = 0 = ⇒ P ( W | Z ) 6 = 0 . P ( X | Z,W ) = P ( X,Z,W ) P ( Z,W ) = P ( X,W | Z ) P ( W | Z ) = ∑ y P ( X,y,W | Z ) P ( W | Z ) By (1): = ∑ y P ( X | Z ) P ( y,W | Z ) P ( W | Z ) = P ( X | Z ) P ( W | Z ) P ( W | Z ) = P ( X | Z ) . (This is in fact the decomposition rule). Hence, P ( X | Z,W ) P ( Y | Z,W ) = P ( X | Z ) P ( Y,W | Z ) P ( W | Z ) By (1): = P ( X,Y,W | Z ) P ( W | Z ) = P ( X,Y | W,Z ) = ⇒ ( X ⊥ Y | Z,W ) 2 (b) Answer: Assume: ( X ⊥ W | Z,Y ) (2) ( X ⊥ Y | Z ) (3) Case analysis: i. Case: P ( Z ) = 0 , or P ( Y,Z ) = 0 , or P ( X,Y,Z ) = 0 = ⇒ simple. ii. Otherwise: P ( X,Y,W | Z ) = P ( X,Y | Z ) P ( W | X,Y,Z ) By (3): = P ( X | Z ) P ( Y | Z ) P ( W | X,Y,Z ) By (2): = P ( X | Z ) P ( Y | Z ) P ( W | Y,Z ) = P ( X | Z ) P ( Y,W | Z ) = ⇒ ( X ⊥ Y,W | Z ) (c) Answer: Assume: ( X...
View Full Document

## This note was uploaded on 02/05/2010 for the course CS CS229 taught by Professor Eransegal during the Fall '07 term at École Normale Supérieure.

### Page1 / 4

ps2_sol - 1 Introduction to Probabilistic Graphical Models...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online