junction tree propagation - bndg 4-4 · 2008. 5. 19. · join tree (junction tree) let g be the set...
TRANSCRIPT
![Page 1: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/1.jpg)
Junction tree propagation - BNDG 4-4.6
Finn V. Jensen and Thomas D. Nielsen
Junction tree propagation – p. 1/25
![Page 2: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/2.jpg)
Exact Inference
Message Passing in Join Trees
More sophisticated inference technique; used in most implemented Bayesian Networksystems (e.g. Hugin).
Overview:
Given: Bayesian network for P (V)
⇓ Preprocessing
Construct a new internal representation for P (V) called a junction tree
⇓ Inference/Updating
Retrieve P (A | E = e) for single A ∈ V
Junction tree propagation – p. 2/25
![Page 3: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/3.jpg)
Calculate a marginal
P (A4)
=X
A1
X
A2
X
A3
X
A5
X
A6
φ1(A1)φ2(A2, A1)φ3(A3, A1)φ4(A4, A2)φ5(A5, A2, A3)φ6(A6, A3)
Junction tree propagation – p. 3/25
![Page 4: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/4.jpg)
Calculate a marginal
P (A4)
=X
A1
X
A2
X
A3
X
A5
X
A6
φ1(A1)φ2(A2, A1)φ3(A3, A1)φ4(A4, A2)φ5(A5, A2, A3)φ6(A6, A3)
=X
A1
φ1(A1)X
A2
φ2(A2, A1)φ4(A4, A2)X
A3
φ3(A3, A1)X
A5
φ5(A5, A2, A3)X
A6
φ6(A6, A3)
Junction tree propagation – p. 3/25
![Page 5: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/5.jpg)
Calculate a marginal
P (A4)=P
A1φ1(A1)
P
A2φ2(A2,A1)φ4(A4,A2)
P
A3φ3(A3,A1)
P
A5φ5(A5,A2,A3)
P
A6φ6(A6,A3)
Junction tree propagation – p. 4/25
![Page 6: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/6.jpg)
Calculate a marginal
Some of the calculationscan be simplified. E.g.
X
A6
P (A6 |A3) = 1
P (A4)=P
A1φ1(A1)
P
A2φ2(A2,A1)φ4(A4,A2)
P
A3φ3(A3,A1)
P
A5φ5(A5,A2,A3)
P
A6φ6(A6,A3)
Junction tree propagation – p. 4/25
![Page 7: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/7.jpg)
Calculate a marginal
Assume evidence e = (e5, e6); represented as 0/1 potentials. Then:
P (A4, e) =X
A1
φ1(A1)X
A2
φ2(A2, A1)φ4(A4, A2)
X
A3
φ3(A3, A1)X
A5
φ5(A5, A2, A3)e5
X
A6
φ6(A6, A3)e6
and
P (A4 | e) =P (A4, e)
P
A4P (A4, e)
Junction tree propagation – p. 5/25
![Page 8: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/8.jpg)
Calculate a marginal
Assume evidence e = (e5, e6); represented as 0/1 potentials. Then:
P (A4, e) =X
A1
φ1(A1)X
A2
φ2(A2, A1)φ4(A4, A2)
X
A3
φ3(A3, A1)X
A5
φ5(A5, A2, A3)e5
X
A6
φ6(A6, A3)e6
and
P (A4 | e) =P (A4, e)
P
A4P (A4, e)
The process is sufficiently general to handle all evidence scenarios!
⇒ We look for a general structure in which these calculations can be performed for allvariables.
Junction tree propagation – p. 5/25
![Page 9: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/9.jpg)
Domain graphs
The Bayesian network The domain graph
The link (A2, A3) is called a moral link.
Junction tree propagation – p. 6/25
![Page 10: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/10.jpg)
Domain graphs
Moralization in general
◮ For all nodes X: Connect pairwise all parents of X with undirected links.
◮ Replace all original directed links by undirected ones.
1 2
3 4
765
8
The Bayesian network
Junction tree propagation – p. 7/25
![Page 11: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/11.jpg)
Domain graphs
Moralization in general
◮ For all nodes X: Connect pairwise all parents of X with undirected links.
◮ Replace all original directed links by undirected ones.
1 2
3 4
765
8
The Bayesian network
1 2
3 4
765
8
Step 1
Junction tree propagation – p. 7/25
![Page 12: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/12.jpg)
Domain graphs
Moralization in general
◮ For all nodes X: Connect pairwise all parents of X with undirected links.
◮ Replace all original directed links by undirected ones.
1 2
3 4
765
8
The Bayesian network
1 2
3 4
765
8
Step 1
1 2
3 4
765
8
Step 2
Junction tree propagation – p. 7/25
![Page 13: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/13.jpg)
Eliminating a variable
Suppose that when calculating P (A4) we start off by eliminating A3:
φ′(A1, A2, A5, A6) =X
A3
φ3(A3, A1)φ4(A4, A2)φ5(A5, A2, A3).
Graphically:
The Bayesian network The domain graph The new domain graph
Junction tree propagation – p. 8/25
![Page 14: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/14.jpg)
Eliminating a variable
Suppose that when calculating P (A4) we start off by eliminating A3:
φ′(A1, A2, A5, A6) =X
A3
φ3(A3, A1)φ4(A4, A2)φ5(A5, A2, A3).
Graphically:
The Bayesian network The domain graph The new domain graph
Perfect elimination sequences
If all variables can be eliminated without introducing fill-in edges, then the eliminationsequence is called perfect. An example could be A5, A6, A3, A1, A2, A4.
Junction tree propagation – p. 8/25
![Page 15: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/15.jpg)
Perfect elimination sequences
Perfect elimination sequences
All perfect elimination sequences produce the same domain set, namely the set of cliques ofthe domain graph.
Clique
A set of nodes is complete if all nodes are pairwise linked. A complete set is a clique if it isnot a subset of another complete set.
1 2
3 4
765
8
Property
Any perfect elimination sequence ending with A is optimal w.r.t. calculating P (A).
Junction tree propagation – p. 9/25
![Page 16: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/16.jpg)
Triangulated graphs
Definition
An undirected graph with a perfect elimination sequence is called a triangulated graph.
Example
not triangulated
triangulated
Junction tree propagation – p. 10/25
![Page 17: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/17.jpg)
Triangulated graphs
Definition
An undirected graph with a perfect elimination sequence is called a triangulated graph.
Example
not triangulated
triangulated
Corollary
A graph is triangulated iff all nodes can successively be eliminated without introducing fill-inedges.
Junction tree propagation – p. 10/25
![Page 18: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/18.jpg)
Triangulated graphs
Definition
An undirected graph with a perfect elimination sequence is called a triangulated graph.
Example
not triangulated
triangulated
Corollary
A graph is triangulated iff there does not exist a cycle of length ≥ 4 that is not cut by a cord.
Junction tree propagation – p. 10/25
![Page 19: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/19.jpg)
Triangulated graphs
Definition
An undirected graph with a perfect elimination sequence is called a triangulated graph.
Example
not triangulated
triangulated
Corollary
In a triangulated graph, each variable A has a perfect elimination sequence ending with A.
Junction tree propagation – p. 10/25
![Page 20: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/20.jpg)
Triangulated graphs
Definition
An undirected graph with a perfect elimination sequence is called a triangulated graph.
Example
not triangulated
triangulated
Assumption
For now we shall assume that the domain graph is triangulated!
Junction tree propagation – p. 10/25
![Page 21: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/21.jpg)
Exact Inference
Join Tree (Junction Tree)
Let G be the set of cliques from an undirected graph, and let the cliques of G be organized ina tree T . Then T is a join tree if
◮ a variable V that is contained in two nodes C,C′ also is contained in every node onthe (unique) path connecting C and C′.
Example:
Junction tree propagation – p. 11/25
![Page 22: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/22.jpg)
Exact Inference
Join Tree (Junction Tree)
Let G be the set of cliques from an undirected graph, and let the cliques of G be organized ina tree T . Then T is a join tree if
◮ a variable V that is contained in two nodes C,C′ also is contained in every node onthe (unique) path connecting C and C′.
Example:
(a) A join tree
(b) Not a join tree
Junction tree propagation – p. 11/25
![Page 23: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/23.jpg)
Exact Inference
Join Tree (Junction Tree)
Let G be the set of cliques from an undirected graph, and let the cliques of G be organized ina tree T . Then T is a join tree if
◮ a variable V that is contained in two nodes C,C′ also is contained in every node onthe (unique) path connecting C and C′.
Example:
(a) A join tree
(b) Not a join tree
Theorem: An undirected graph G is triangulated if and only if the cliques of G can beorganized in a join tree.
Junction tree propagation – p. 11/25
![Page 24: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/24.jpg)
Join tree construction
From undirected graph to join graph
Define undirected graph (C, E∗):
C : Set of cliques in triangulated moral graphE∗ : {(C1,C2) | C1 ∩ C2 6= ∅}
Junction tree propagation – p. 12/25
![Page 25: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/25.jpg)
Join tree construction
From undirected graph to join graph
Define undirected graph (C, E∗):
C : Set of cliques in triangulated moral graphE∗ : {(C1,C2) | C1 ∩ C2 6= ∅}
1 2
3 4
765
8
Junction tree propagation – p. 12/25
![Page 26: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/26.jpg)
Join tree construction
From undirected graph to join graph
Define undirected graph (C, E∗):
C : Set of cliques in triangulated moral graphE∗ : {(C1,C2) | C1 ∩ C2 6= ∅}
1 2
3 4
765
8
Junction tree propagation – p. 12/25
![Page 27: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/27.jpg)
Join tree construction
From undirected graph to join graph
Define undirected graph (C, E∗):
C : Set of cliques in triangulated moral graphE∗ : {(C1,C2) | C1 ∩ C2 6= ∅}
1 2
3 4
765
8
Junction tree propagation – p. 12/25
![Page 28: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/28.jpg)
Join tree construction
From undirected graph to join graph
Define undirected graph (C, E∗):
C : Set of cliques in triangulated moral graphE∗ : {(C1,C2) | C1 ∩ C2 6= ∅}
1 2
3 4
765
8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
Junction tree propagation – p. 12/25
![Page 29: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/29.jpg)
Join tree construction
From join graph to join tree
Find maximal spanning tree: for edge e = (C1,C2) define weight w(e) := |C1 ∩ C2|. Let(C, E) be a maximal spanning tree of (C, E∗) with respect to edge weights.
Junction tree propagation – p. 13/25
![Page 30: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/30.jpg)
Join tree construction
From join graph to join tree
Find maximal spanning tree: for edge e = (C1,C2) define weight w(e) := |C1 ∩ C2|. Let(C, E) be a maximal spanning tree of (C, E∗) with respect to edge weights.
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
Junction tree propagation – p. 13/25
![Page 31: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/31.jpg)
Join tree construction
From join graph to join tree
Find maximal spanning tree: for edge e = (C1,C2) define weight w(e) := |C1 ∩ C2|. Let(C, E) be a maximal spanning tree of (C, E∗) with respect to edge weights.
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1
1
1
2 2
2
2
Junction tree propagation – p. 13/25
![Page 32: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/32.jpg)
Join tree construction
From join graph to join tree
Find maximal spanning tree: for edge e = (C1,C2) define weight w(e) := |C1 ∩ C2|. Let(C, E) be a maximal spanning tree of (C, E∗) with respect to edge weights.
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1
1
1
2 2
2
2
Junction tree propagation – p. 13/25
![Page 33: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/33.jpg)
Join tree construction
From join graph to join tree
Find maximal spanning tree: for edge e = (C1,C2) define weight w(e) := |C1 ∩ C2|. Let(C, E) be a maximal spanning tree of (C, E∗) with respect to edge weights.
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
Junction tree propagation – p. 13/25
![Page 34: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/34.jpg)
Join tree construction
Another example
Junction tree propagation – p. 14/25
![Page 35: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/35.jpg)
Join tree construction
Another example
A1
A2 A3
A4 A5 A6
Junction tree propagation – p. 14/25
![Page 36: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/36.jpg)
Join tree construction
Another example
A1
A2 A3
A4 A5 A6
11
1 1
2A2, A4
A1, A2, A3
A2, A3, A5
A2, A3, A5
Junction tree propagation – p. 14/25
![Page 37: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/37.jpg)
Join tree construction
Another example
A1
A2 A3
A4 A5 A6
11
1 1
2A2, A4
A1, A2, A3
A2, A3, A5
A2, A3, A5
A2, A4
A1, A2, A3
A2, A3, A5
A2, A3, A5
Junction tree propagation – p. 14/25
![Page 38: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/38.jpg)
Junction tree construction
Correctness
Let (V,→) be a triangulated graph, (C, E∗). Every maximal spanning tree (C, E) of (C, E∗) isa join tree for V (Jensen,Jensen 1994).
If (V,→) is the triangulated moral graph of a Bayesian network, then there exists for everyV ∈ V a clique C ∈ C with {V } ∪ pa(V ) ⊆ C.
Junction tree propagation – p. 15/25
![Page 39: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/39.jpg)
Propagation in junction trees
Initialization
Let T be a join tree for the domain graph G with potentials Φ. A junction tree for G consistsof T with the additions:
◮ each potential φ from Φ is assigned to a clique containing dom (φ).
◮ each link has a separator attached.
◮ each separator contains two mailboxes, one for each direction.
Example
Junction tree propagation – p. 16/25
![Page 40: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/40.jpg)
Propagation in junction trees
Propagation
To calculate P (A4) we find a clique (V6) containing A4 and send messages to that clique.Example
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
↓
φ4 φ5 φ6
↑ ↑ ↑ψ4
ψ2 = φ↓S2
5 ψ1 = φ↓S1
6
Junction tree propagation – p. 17/25
![Page 41: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/41.jpg)
Propagation in junction trees
Propagation
To calculate P (A4) we find a clique (V6) containing A4 and send messages to that clique.Example
1 : ψ1 = φ↓S1
6
↑1 : ψ2 = φ↓S2
5
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
↓
φ4 φ5 φ6
↑ ↑
Junction tree propagation – p. 17/25
![Page 42: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/42.jpg)
Propagation in junction trees
Propagation
To calculate P (A4) we find a clique (V6) containing A4 and send messages to that clique.Example
↓
↑ ↑ ↑1 : ψ1 = φ↓S1
61 : ψ2 = φ↓S2
5
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
φ4 φ5 φ6
2 : ψ4 = . . .
ψ4 =P
A1φ1φ2
P
A3φ3ψ1ψ2
Now evidence has been collected to V4 and we get P (A4) =P
A2φ4ψ4.
Junction tree propagation – p. 17/25
![Page 43: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/43.jpg)
Propagation in junction trees
Propagation
To calculate P (Ai) for any Ai we send messages away from V6.
Example continued
↓
↑ ↑ ↑1 : ψ1 = φ↓S1
61 : ψ2 = φ↓S2
5
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
φ4 φ5 φ6
2 : ψ4 = . . .
ψ4 =P
A1φ1φ2
P
A3φ3ψ1ψ2
Junction tree propagation – p. 18/25
![Page 44: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/44.jpg)
Propagation in junction trees
Propagation
To calculate P (Ai) for any Ai we send messages away from V6.
Example continued
↓
φ6
↑ ↑ ↑1 : ψ1 = φ↓S1
61 : ψ2 = φ↓S2
5
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
φ4 φ5
2 : ψ4 = . . .
3 : ψ4 = φ↓A2
4
Junction tree propagation – p. 18/25
![Page 45: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/45.jpg)
Propagation in junction trees
Propagation
To calculate P (Ai) for any Ai we send messages away from V6.
Example continued
φ5
V6 : A2, A4 V2 : A2, A3, A5 V1 : A3, A6
φ1, φ2, φ3
↓
φ4 φ6
↑ ↑ ↑1 : ψ1 = φ↓S1
61 : ψ2 = φ↓S2
5
V4 : A1, A2, A3
↓ ↓
S4 : A2 S2 : A2, A3 S1 : A3
4 : ψ2 = . . . 4 : ψ1 = . . .
ψ2 =P
A1ψ1φ1φ2φ3ψ4 ∼ {ψ4,
P
A1φ1φ2φ3, ψ1}
ψ1 =P
A2ψ2ψ4
P
A1φ1φ2φ3
2 : ψ4 = . . .
3 : ψ4 = φ↓A2
4
Junction tree propagation – p. 18/25
![Page 46: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/46.jpg)
Propagation in junction trees
In general: Sending a message from Ci to Cj
CiCj
Cl
Cm
mij
mli
mmi
mij :=X
Cj\Ci
Y
φ∈ΦCi
φY
Ck∈nd(Ci)
mki
Junction tree propagation – p. 19/25
![Page 47: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/47.jpg)
Exact Inference
Messages passing in general
R
Collect and distribute messages
i) Collect messages to a preselected root R
Junction tree propagation – p. 20/25
![Page 48: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/48.jpg)
Exact Inference
Messages passing in general
R
1
11
Collect and distribute messages
i) Collect messages to a preselected root R
Junction tree propagation – p. 20/25
![Page 49: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/49.jpg)
Exact Inference
Messages passing in general
R2
1
1
2
1
Collect and distribute messages
i) Collect messages to a preselected root R
Junction tree propagation – p. 20/25
![Page 50: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/50.jpg)
Exact Inference
Messages passing in general
R2
1
1
2
1
3
3
Collect and distribute messages
i) Collect messages to a preselected root R
ii) Distribute messages away from the root R
Junction tree propagation – p. 20/25
![Page 51: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/51.jpg)
Exact Inference
Messages passing in general
42
1
1
2
1
R
3
4
3
4
Collect and distribute messages
i) Collect messages to a preselected root R
ii) Distribute messages away from the root R
Junction tree propagation – p. 20/25
![Page 52: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/52.jpg)
Propagation in junction trees
Theorem
Let the junction tree T represent the Bayesian network BN over the universe U and withevidence e. Assume that T is full.
1. Let V be a clique with set of potentials ΦV , and let S1, . . . , Sk be V ’s neighboringseparators and with V -directed messages Ψ1, . . . ,Ψk. Then
P (V, e) =Y
ΦV
Y
Ψ1 · . . . ·Y
Ψk.
2. Let S be a separator with the sets ΨS and ΨS in the mailboxes. Then
P (S, e) =Y
ΨS
Y
ΨS .
Junction tree propagation – p. 21/25
![Page 53: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/53.jpg)
Propagation in junction trees
Theorem
Let the junction tree T represent the Bayesian network BN over the universe U and withevidence e. Assume that T is full.
1. Let V be a clique with set of potentials ΦV , and let S1, . . . , Sk be V ’s neighboringseparators and with V -directed messages Ψ1, . . . ,Ψk. Then
P (V, e) =Y
ΦV
Y
Ψ1 · . . . ·Y
Ψk.
2. Let S be a separator with the sets ΨS and ΨS in the mailboxes. Then
P (S, e) =Y
ΨS
Y
ΨS .
Evidence
Evidence is inserted by adding corresponding 0/1-potentials to the appropriate cliques.
Junction tree propagation – p. 21/25
![Page 54: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/54.jpg)
Triangulation of graphs
So far we have assumed that the domain graph is triangulated. If this is not the case, thenwe embed it in a triangulated graph.
Example
Junction tree propagation – p. 22/25
![Page 55: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/55.jpg)
Triangulation of graphs
So far we have assumed that the domain graph is triangulated. If this is not the case, thenwe embed it in a triangulated graph.
Example
A triangulated graph extending (b)
Junction tree propagation – p. 22/25
![Page 56: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/56.jpg)
Triangulation of graphs
1 2
3 4
765
8
Junction tree propagation – p. 23/25
![Page 57: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/57.jpg)
Triangulation of graphs
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 58: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/58.jpg)
Triangulation of graphs
1 2
3 4
765
8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1
1
1
2 2
2
2
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 59: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/59.jpg)
Triangulation of graphs
1 2
3 4
765
8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1
1
1
2 2
2
2
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 60: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/60.jpg)
Triangulation of graphs
1 2
3 4
765
8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 61: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/61.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 62: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/62.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 63: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/63.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 64: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/64.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 65: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/65.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
Junction tree propagation – p. 23/25
![Page 66: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/66.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
A heuristic
Repeatedly eliminate a simplicial node, and if this is not possible eliminate a node minimizing(fa(X) are noneliminated neighbors of X):
◮ |fa(X)| (fill-in-size ).
◮ | sp(fa(X)) | (clique-size).
◮P
{Y,Z}:{Y,Z}∈nb(X)∧Z 6∈nb(Y ) | sp({Y, Z}) | (fill-in-weight)
Junction tree propagation – p. 23/25
![Page 67: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/67.jpg)
Triangulation of graphs
1 2
3 4
765
8
1 2
3 4
765
8
Usually there are many possible triangulations and finding the best one is NP-hard!
Junction tree propagation – p. 23/25
![Page 68: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/68.jpg)
Summary
1 2
3 4
765
8
Bayesian network
Junction tree propagation – p. 24/25
![Page 69: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/69.jpg)
Summary
1 2
3 4
765
8
1 2
3 4
765
8
Bayesian network Moral graph
Junction tree propagation – p. 24/25
![Page 70: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/70.jpg)
Summary
1 2
3 4
765
8
1 2
3 4
765
8
1 2
3 4
765
8
Bayesian network Moral graph Triangulated graph
Junction tree propagation – p. 24/25
![Page 71: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/71.jpg)
Summary
1 2
3 4
765
8
1 2
3 4
765
8
1 2
3 4
765
8
Bayesian network Moral graph Triangulated graph
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
Join graph
Junction tree propagation – p. 24/25
![Page 72: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/72.jpg)
Summary
1 2
3 4
765
8
1 2
3 4
765
8
1 2
3 4
765
8
Bayesian network Moral graph Triangulated graph
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
Join graph Join tree
Junction tree propagation – p. 24/25
![Page 73: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/73.jpg)
Summary
1 2
3 4
765
8
1 2
3 4
765
8
1 2
3 4
765
8
Bayesian network Moral graph Triangulated graph
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
6,7
1,2,3 2,4,7
3,6,72,3,7
5,6,7,8
2,3 2,7
3,7
Join graph Join tree Junction tree
Junction tree propagation – p. 24/25
![Page 74: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/74.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
Junction tree propagation – p. 25/25
![Page 75: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/75.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
⇓ Moralize
Domain graph G for BN (also called the moral graph)
Junction tree propagation – p. 25/25
![Page 76: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/76.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
⇓ Moralize
Domain graph G for BN (also called the moral graph)
⇓ Triangulate
Triangulated graph for G
Junction tree propagation – p. 25/25
![Page 77: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/77.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
⇓ Moralize
Domain graph G for BN (also called the moral graph)
⇓ Triangulate
Triangulated graph for G
⇓ Find cliques
Join graph
Junction tree propagation – p. 25/25
![Page 78: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/78.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
⇓ Moralize
Domain graph G for BN (also called the moral graph)
⇓ Triangulate
Triangulated graph for G
⇓ Find cliques
Join graph
⇓ Find maximal spanning tree
Join tree
Junction tree propagation – p. 25/25
![Page 79: Junction tree propagation - BNDG 4-4 · 2008. 5. 19. · Join Tree (Junction Tree) Let G be the set of cliques from an undirected graph, and let the cliques of G be organized in a](https://reader033.vdocument.in/reader033/viewer/2022060914/60a7faadd79f715ad65b87bf/html5/thumbnails/79.jpg)
Exact Inference
Summary
Given: Bayesian network BN for P (V)
⇓ Moralize
Domain graph G for BN (also called the moral graph)
⇓ Triangulate
Triangulated graph for G
⇓ Find cliques
Join graph
⇓ Find maximal spanning tree
Join tree
⇓
Perform belief updating by message passing.
Junction tree propagation – p. 25/25