RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability...
Transcript of RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability...
![Page 1: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/1.jpg)
Bayesian Belief Networks
RN, Chapter 14
![Page 2: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/2.jpg)
2
Decision Theoretic Agents
Introduction to Probability [Ch13]
Belief networks [Ch14]Introduction [Ch14.1-14.2]
Bayesian Net Inference [Ch14.4](Bucket Elimination)
Dynamic Belief Networks [Ch15]
Single Decision [Ch16]
Sequential Decisions [Ch17]
![Page 3: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/3.jpg)
3
![Page 4: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/4.jpg)
4
Motivation
Gates says [LATimes, 28/Oct/96]:
Microsoft’s competitive advantages is its expertise in “Bayesian networks”
Current ProductsMicrosoft Pregnancy and Child Care (MSN)Answer Wizard (Office, …)Print Troubleshooter
Excel Workbook TroubleshooterOffice 95 Setup Media TroubleshooterWindows NT 4.0 Video TroubleshooterWord Mail Merge Troubleshooter
![Page 5: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/5.jpg)
5
Motivation (II)
US Army: SAIP (Battalion Detection from SAR, IR… GulfWar)
NASA: Vista (DSS for Space Shuttle)GE: Gems (real-time monitor for utility generators)Intel: (infer possible processing problems from end-of-line tests on
semiconductor chips)
KIC: medical: sleep disorders, pathology, trauma care, hand and wrist evaluations, dermatology, home-based health evaluationsDSS for capital equipment: locomotives, gas-turbine engines, office equipment
![Page 6: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/6.jpg)
6
Motivation (III)
Lymph-node pathology diagnosisManufacturing controlSoftware diagnosisInformation retrievalTypes of tasks
Classification/RegressionSensor FusionPrediction/ForecastingModeling
![Page 7: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/7.jpg)
7
MotivationChallenge: To decide on proper action
Which treatment, given symptoms?Where to move?Where to search for info?. . .
Need to know dependencies in worldbetween symptom and diseasebetween symptom1 and symptom2
between disease1 and disease2
. . .Q: Full joint?
A: Too big (≥ 2n)Too slow (inference requires adding 2k . . . )
Better:Encode dependenciesEncode only relevant dependencies
![Page 8: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/8.jpg)
8
Components of a Bayesian Net
Nodes: one for each random variableArcs: one for each direct influence between two random variablesCPT: each node stores a conditional probability table
P( Node | Parents(Node) )to quantify effects of “parents" on child
![Page 9: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/9.jpg)
9
Causes, and Bayesian NetWhat “causes” Alarm?A: Burglary, Earthquake
What “causes” JohnCall?A: AlarmN.b., NOT Burglary, ...
Why not Alarm ⇒ MaryCalls?
A: Mary not always home... phone may be broken...
![Page 10: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/10.jpg)
10
Independence in a Belief Net
Burglary, Earthquakeindependent
B ⊥ E
Given Alarm,JohnCalls
and MaryCalls
independent
J ⊥ M | A
JohnCalls is correlated with MaryCalls ¬(J ⊥ M )as suggest AlarmBut given Alarm,JohnCalls
gives no NEW evidence wrt
MaryCalls
![Page 11: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/11.jpg)
11
Conditional Independence
B ⊥ E | {} (B ⊥ E)M ⊥ {B,E,J} | AGiven graph G, ILM(G) = { (Xi ⊥ NonDescendantsXi | PaXi) }
Local Markov Assumption:A variable X is independentof its non-descendants given its parents (Xi ⊥
NonDescendantsXi | PaXi )
![Page 12: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/12.jpg)
13
Factoid: Chain Rule
P(A,B,C) = P(A | B,C) P(B,C)
= P(A | B,C) P(B|C) P(C)In general:P(X1,X2, ... ,Xm ) =P(X1
| X2
, ... ,Xm
) P(X2
, ... ,Xm
) =P(X1
| X2
, ... ,Xm
) P(X2
| X3
, ... ,Xm
) P( X3
, ... ,Xm
) =
∏i
P(Xi
| Xi+1
, ... ,Xm
)
![Page 13: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/13.jpg)
14
Joint Distribution
P( +j, +m, +a, -b, -e )= P( +j | +m, +a, -b, -e )
P(+m
| +a, -b, -e )
P(+a| -b, -e )
P(-b | -e )
P(-e )
P( +j | +a )J ⊥
{M,B,E} | A
P( +m | +a )M ⊥
{B,E} | A
P( +a | -b,-e )
P(-b)B ⊥
E
P(-e )
![Page 14: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/14.jpg)
15
Joint Distribution
P( +j, +m, +a, -b, -e )= P( +j | +a)
P(+m
| +a)
P(+a| -b, -e )
P(-b)
P(-e )
![Page 15: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/15.jpg)
16
Recovering Joint
![Page 16: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/16.jpg)
17
Meaning of Belief NetA BN represents
joint distributioncondition independence statements
P( J, M, A, ¬B, ¬E )= P(¬B ) P(¬E ) P(A|¬B, ¬E) P( J | A) P(M |A)= 0.999 ×
0.998 ×
0.001 ×
0.90 ×
0.70 = 0.00062
In gen'l, P(X1,X2, . . . ,Xm ) = ∏i P(Xi |Xi+1, . . . ,Xm )Independence means
P(Xi
|Xi+1
, . . . ,Xm
) = P(Xi
| Parents(Xi
) )Node independent of predecessors, given parents
So... P(X1,X2, . . . ,Xm ) = ∏i P(Xi | Parents(Xi) )
![Page 17: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/17.jpg)
18
Comments
BN used 10 entries... can recover full joint (25
entries)
(Given structure, other 25
–
10 entries are REDUNDANT)
⇒
Can computeP( Burglary | JohnCalls, ¬MaryCalls
)
:
Get joint, then marginalize, conditionalize, ...∃ better ways. . .Note: Given structure, ANY CPT is consistent.∄
redundancies in BN. . .
![Page 18: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/18.jpg)
19
Conditional IndependenceNode X is independent of its non-descendantsgiven assignment to immediate parents parents(X)
General question: “X ⊥ Y | E”Are nodes X independent of nodes Y,given assignments to (evidence) nodes E?
Answer: If every undirected path from X to Yis d-separated by E, then X ⊥
Y
| E
d-separated if every path from X to Y is blocked by E. . . if ∃
node Z
on path s.t.
1.
Z ∈
E, and Z has 1 out-link (on path)2.
Z ∈
E, and Z
has 2 out-link, or3.
Z
has 2 in-links, Z ∉
E, no child of Z
in
E
![Page 19: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/19.jpg)
20
d-separation Conditions
X Z Y
X ⊥
Y | Z
X ⊥
Y | Z
¬(X ⊥
Y | Z)
¬(X ⊥
Y)
¬(X ⊥
Y)
X ⊥
Y
X Z Y
X Z Y
Z
Z
Z
![Page 20: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/20.jpg)
21
d-Separation
Burglary and JohnCalls areconditionally independent given AlarmJohnCalls and MaryCalls areconditionally independent given AlarmBurglary and Earthquake areindependent given no other informationBut. . .
Burglary and Earthquake are dependent given AlarmIe, Earthquake may “explain away” Alarm… decreasing prob of Burglary
![Page 21: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/21.jpg)
22
“V"-Connections
What colour are my wife's eyes?Would it help to know MY eye color?NO! H_Eye
and W_Eye
are independent!
We have a DAUGHTER, who has BLUE eyesNow do you want to know my eye-color?
H_Eye and W_Eye became dependent!
![Page 22: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/22.jpg)
24
Example of d-separation, IId-separated if
every path from X
to Y
is blocked by E
Is Radio
d-separated from Gas
given . . .1. E = {}
?YES: P(R | G ) =
P( R )Starts
∉
E, and Starts
has 2 in-links 2. E = Starts
?NO!! P(R | G, S
) ≠
P(R| S)Starts
∈
E, and Starts
has 2 in-links3. E = Moves
?NO!! P(R | G, M
) ≠
P(R| M)Moves
∈
E, Moves
child-of Starts, and Starts
has 2 in-links (on path)4. E = SparkPlug
?YES: P(R | G, Sp
) =
P(R| Sp)SparkPlug
∈
E, and SparkPlug
has 1 out-link5. E = Battery
?YES: P(R | G, B
) =
P(R| B)Battery
∈
E, and Battery
has 2 out-links
If car does not start, expect radio to NOT work.
Unless you see it is out of gas!If car does not MOVE, expect radio to NOT work.
Unless you see it is out of gas!
![Page 23: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/23.jpg)
25
Markov Blanket
Each node is conditionally independent of all others
given its Markov blanket:parentschildrenchildren's parents
![Page 24: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/24.jpg)
26
Simple Forms of CPTableIn gen'l: CPTable is function mapping
values of parents to distribution over child
Standard: Include ∏U∈ Parents(X)|Dom(U)| rows,each with |Dom(X)| - 1 entriesBut... can be structure within CPTable:Deterministic, Noisy-Or, Decision Tree, . . .
f( +Col, -Flu, +Mal ) = ⟨0.94 0.06⟩
![Page 25: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/25.jpg)
27
Deterministic Node
Given value of parent(s),specify unique value for child(logical, functional)
![Page 26: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/26.jpg)
28
Noisy-OR CPTable
Each cause is independent of the othersAll possible causes are listed
Want: No Fever
if none of Cold, Flu
or MalariaP( ¬Fev
| ¬Col, ¬Flu, ¬Mal ) = 1.0
+
Whatever inhibits cold
from causing feveris independent of
whatever inhibits flu
from causing feverP(¬Fev
| Cold, Flu ) ≈
P(¬Fev
| Cold ) ×
P(¬Fev
| Flu )
Fever
Cold Flu Malaria
![Page 27: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/27.jpg)
29
Noisy-OR “CPTable” (2)
0.6 0.20.1
Fever
Cold Flu Malaria
![Page 28: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/28.jpg)
30
Noisy-Or … expanded
Fever
Cold’ Flu’ Malaria’
Cold Flu Malaria
c f m P(+Fever|c, f, m)
+ + + 1.0
+ + – 1.0
+ – + 1.0
+ – – 1.0
+ + + 1.0
– + – 1.0
– – + 1.0
– – – 0.0
c P(+cold’
| c) P(-cold’
| c)
+ 1-qc
= 0.4 qc
= 0.6
– 0.0 1.0
0.6 0.2 0.1
![Page 29: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/29.jpg)
31
Noisy-Or (Gen'l)
CPCS Network:• Modeling disease/symptom for internal medicine
• Using Noisy-Or & Noisy-Max
• 448 nodes, 906 links
• Required 8,254 values (not 13,931,430) !
![Page 30: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/30.jpg)
32
DecisionTree CPTable
![Page 31: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/31.jpg)
33
Hybrid (discrete+continuous)
Networks
Discrete: Subsidy?, Buys?Continuous: Harvest, Cost
Option 1: Discretizationbut possibly large errors, large CPTs
Option 2: Finitely parameterized canonical familiesProblematic cases to consider. . .
Continuous variable, discrete+continuous parentsCostDiscrete variable, continuous parentsBuys?
![Page 32: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/32.jpg)
34
Continuous Child VariablesFor each “continuous” child E,
with continuous parents Cwith discrete parents D
Need conditional density functionP(E = e | C = c, D = d ) = PD=d
(E
= e | C = c)for each assignment to discrete parents D=dCommon: linear Gaussian model
Need parameters:σt
at btσf af
bf
f( Harvest, Subsidy? ) = “dist over Cost”
![Page 33: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/33.jpg)
35
If everything is Gaussian...
All nodes continuous w/ LG dist'ns⇒
full joint is a multivariate Gaussian
Discrete+continuous LG network⇒
conditional Gaussian network
multivariate Gaussian over all continuous variables for each combination of discrete variable values
![Page 34: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/34.jpg)
36
Discrete variable w/ Continuous Parents
Probability of Buys? given Cost≈? “soft”
threshold:
Probit distribution uses integral of Gaussian:
≈
hard threshold, whose location is subject to noise
![Page 35: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/35.jpg)
37
Logit
vs
Probit
![Page 36: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/36.jpg)
38
Example: Car Diagnosis
![Page 37: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/37.jpg)
39
MammoNet
![Page 38: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/38.jpg)
40
ALARM
A Logical Alarm Reduction Mechanism• 8 diagnoses, 16 findings, …
![Page 39: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/39.jpg)
41
Troup Detection
![Page 40: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/40.jpg)
42
ARCO1: Forecasting Oil Prices
![Page 41: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/41.jpg)
43
ARCO1: Forecasting Oil Prices
![Page 42: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/42.jpg)
44
Forecasting Potato Production
![Page 43: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/43.jpg)
45
Warning System
![Page 44: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/44.jpg)
46
Uses of Belief Nets #1Medical Diagnosis: “Assist/Critique” MD
identify diseases not ruled-outspecify additional tests to performsuggest treatments appropriate/cost-effectivereact to MD’s proposed treatment
Decision Support: Find/repair faults in complex machines[Device, or Manufacturing Plant, or …]… based on sensors, recorded info, history,…
Preventative Maintenance: Anticipate problems in complex machines
[Device, or Manufacturing Plant, or …]…based on sensors, statistics, recorded info, device history,…
![Page 45: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/45.jpg)
47
Uses (con’t)
Logistics Support: Stock warehouses appropriately…based on (estimated) freq. of needs, costs,Diagnose Software:
Find most probable bugs, givenprogram behavior, core dump, source code, …
Part Inspection/Classification:… based on multiple sensors, background, model of production,…Information Retrieval:
Combine information from various sources,based on info from various “agents”,…
General: Partial Info, Sensor fusion-Classification -Interpretation-Prediction -…
![Page 46: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/46.jpg)
48
Belief Nets vs
RulesBoth have “Locality”
Specific clusters (rules / connected nodes)
WHY?: Easier for people to reason CAUSALLYeven if use is DIAGNOSTIC
BN provide OPTIMAL way to deal with+ Uncertainty+ Vagueness (var not given, or only dist)+ Error
…Signals meeting Symbols …
BN permits different “direction”s of inference
Often same nodes (rep’ning Propositions) butBN: Cause ⇒
Effect
“Hep ⇒ Jaundice” P(J | H )
Rule: Effect ⇒
Cause“Jaundice ⇒
Hep”
![Page 47: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/47.jpg)
49
Belief Nets vs
Neural Nets
Both have “graph structure” but
So harder toInitialize NNExplain NN
(But perhaps easier to learn NN from examples only?)
BNs can deal withPartial InformationDifferent “direction”s of inference
BN: Nodes have SEMANTICsCombination Rules: Sound Probability
NN: Nodes: arbitraryCombination Rules: Arbitrary
![Page 48: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/48.jpg)
50
Belief Nets vs
Markov NetsEach uses “graph structure”
to FACTOR a distribution…
explicitly specify dependencies, implicitly independencies…
but subtle differences…BNs capture “causality”, “hierarchies”MNs capture “temporality”
C
BATechnical: BNs use DIRECTRED arcs⇒
allow “induced dependencies”
I (A, {}, B) “A independent of B, given {}”¬ I (A, C, B) “A dependent on B, given C”
MNs use UNDIRECTED arcs ⇒
allow other independencies
I(A, BC, D) A independent of D, given B, CI(B, AD, C) B independent of C, given A, D
D
CB
A
![Page 49: RN, Ch apt er 14rgreiner/C-366/... · 2 Decision Theoretic Agents Introduction to Probability [Ch13] Belief networks [Ch14] Introduction [Ch14.1-14.2] Bayesian Net Inference [Ch14.4]](https://reader035.fdocuments.us/reader035/viewer/2022071211/60229483e728ad72415a4bc3/html5/thumbnails/49.jpg)
51
Summary Components of Belief NetConditional Independenced-separation
V-connectionsMarkov blanket
CPtablesSpecial casesContinuous
Deployed ExamplesComparison to other Rep’ns