CS 3710 Advanced Topics in AI

Lecture 2

Probabilistic graphical models

Milos Hauskrecht
[email protected]
5329 Sennott Square, x4-8845

http://www.cs.pitt.edu/~milos/courses/cs3710/

CS 3710 Probabilistic Graphical Models

Motivation. Medical example.

We want to build a system for the diagnosis of pneumonia.
Problem description:
Disease: pneumonia
Patient symptoms (findings, lab tests):
Fever, Cough, Paleness, WBC (white blood cells) count,
Chest pain, etc.
Representation of a patient case:
Statements that hold (are true) for the patient.
E.g: Fever =True
Cough =False
WBCcount=High
Diagnostic task: we want to decide whether the patient suffers
from the pneumonia or not given the symptoms
CS 3710 Probabilistic Graphical Models

1
 Uncertainty
To make diagnostic inference possible we need to represent
knowledge (axioms) that relate symptoms and diagnosis

Pneumonia

Paleness Fever Cough WBC count

Problem: disease/symptoms relations are not deterministic

They are uncertain (or stochastic) and vary from patient
to patient

CS 3710 Probabilistic Graphical Models

Modeling the uncertainty.

Key challenges:
How to represent uncertain relations?
How to manipulate such knowledge to make inferences?
Humans can reason with uncertainty.

Pneumonia

Paleness Fever Cough WBC count

CS 3710 Probabilistic Graphical Models

2
 Modeling uncertainty with probabilities
Random variables:
Binary Pneumonia is either True, False
Random variable Values
Multi-valued Pain is one of {Nopain, Mild, Moderate, Severe}
Random variable Values
Continuous
HeartRate is a value in < 0 ; 250 >
Random variable Values

A multivariate random variable or random vector is a

vector whose components are individual random variables
A patient state: an assignment of values to random
variables. A value of a multivariate random var.
E.g. Pneumonia =T , Fever =T, Paleness=F,
WBCcount=medium, Cough=False
CS 3710 Probabilistic Graphical Models

Probabilities
Quantifies how likely is the outcome of a random variable
Unconditional probabilities (prior probabilities)

P( Pneumonia = True) = 0.001

P( Pneumonia = False) = 0.999
P(WBCcount = high) = 0.005
Probability distribution
Defines probabilities for all possible value assignments to a
random variable
Pneumonia P(Pneumonia)
Values are mutually exclusive
True 0.001
False 0.999

CS 3710 Probabilistic Graphical Models

3
 Probability distribution
Defines probability for all possible value assignments
Example 1:
Pneumonia P(Pneumonia)
P( Pneumonia = True) = 0.001
True 0.001
P( Pneumonia = False) = 0.999 False 0.999
P( Pneumonia = True) + P( Pneumonia = False) = 1
Probabilities sum to 1 !!!

Example 2:
P(WBCcount = high) = 0.005 WBCcount P(WBCcount)
P(WBCcount = normal) = 0.993 high 0.005
normal 0.993
P(WBCcount = high) = 0.002 low 0.002

CS 3710 Probabilistic Graphical Models

Joint probability distribution

Joint probability distribution (for a set variables)
Defines probabilities for all possible assignments of values to
variables in the set
Example: variables Pneumonia and WBCcount
P ( pneumonia , WBCcount )
Is represented by 2 3 matrix

WBCcount
high normal low

Pneumonia True 0.0008 0.0001 0.0001

False 0.0042 0.9929 0.0019

CS 3710 Probabilistic Graphical Models

4
 Joint probabilities
Marginalization
reduces the dimension of the joint distribution
Sums variables out
P ( pneumonia , WBCcount ) 2 3 matrix
P(Pneumonia)
WBCcount
high normal low

Pneumonia True 0.0008 0.0001 0.0001 0.001

False 0.0042 0.9929 0.0019 0.999
0.005 0.993 0.002

P(WBCcount)
Marginalization (here summing of columns or rows)
CS 3710 Probabilistic Graphical Models

Full joint distribution

the joint distribution for all variables in the problem
It defines the complete probability model for the problem

Example: pneumonia diagnosis

Variables: Pneumonia, Fever, Paleness, WBCcount, Cough
Full joint defines the probability for all possible assignments of
values to Pneumonia, Fever, Paleness, WBCcount, Cough

P( Pneumonia= T ,WBCcount= High, Fever = T , Cough = T , Paleness= T )

P( Pneumonia= T ,WBCcount= High, Fever = T , Cough = T , Paleness= F )
P( Pneumonia= T ,WBCcount = High, Fever = T , Cough = F , Paleness = T )
K etc
CS 3710 Probabilistic Graphical Models

5
 Conditional probabilities
Conditional probability distribution
Defines probabilities of outcomes of a variable, given a fixed
assignment to some other variable values
P( Pneumonia = true | WBCcount = high)

P( Pneumonia | WBCcount) 3 element vector of 2 elements

WBCcount
high normal low

Pneumonia True 0.08 0.0001 0.0001

False 0.92 0.9999 0.9999
1.0 1.0 1.0
P( Pneumonia = true | WBCcount = high)
+ P( Pneumonia = false | WBCcount = high)
CS 3710 Probabilistic Graphical Models

Conditional probabilities

Conditional probability
Is defined in terms of the joint probability:
P( A, B)
P( A | B) = s.t. P( B) 0
P( B)
Example:
P( pneumonia = true | WBCcount = high) =
P( pneumonia= true,WBCcount = high)
P(WBCcount = high)

P( pneumonia = false | WBCcount = high) =

P( pneumonia= false,WBCcount = high)
P(WBCcount = high)
CS 3710 Probabilistic Graphical Models

6
 Conditional probabilities
Conditional probability distribution.
P( A, B)
P( A | B) = s.t. P( B) 0
P( B)
Product rule. Joint probability can be expressed in terms of
conditional probabilities

P ( A, B) = P( A | B) P( B)

Chain rule. Any joint probability can be expressed as a

product of conditionals
P( X 1 , X 2 ,K X n ) = P( X n | X 1, K X n 1 ) P( X 1, K X n 1 )
= P( X n | X 1, K X n 1 ) P ( X n 1 | X 1, K X n 2 ) P( X 1, K X n 2 )
= i =1 P( X i | X 1, K X i 1 )
n

CS 3710 Probabilistic Graphical Models

Bayes rule
Conditional probability.
P( A, B) P( A, B) = P( B | A) P( A)
P( A | B) =
P( B)
Bayes rule: P( B | A) P( A)
P( A | B) =
P( B)
When is it useful?
When we are interested in computing the diagnostic query
from the causal probability
P(effect | cause) P(cause)
P(cause | effect ) =
P(effect )
Reason: It is often easier to assess causal probability
E.g. Probability of pneumonia causing fever
vs. probability of pneumonia given fever

CS 3710 Probabilistic Graphical Models

7
 Bayes rule
Assume a variable A with multiple values a1 , a2 ,K ak
Bayes rule can be rewritten as:

P(B = b | A = a j )P( A = a j )
P ( A = a j | B = b) =
P ( B = b)
P(B = b | A = a j )P( A = a j )
=

k
i =1
P(B = b | A = a j )P( A = a j )

Used in practice when we want to compute:

P ( A | B = b) for all values of a1 , a2 ,K ak

CS 3710 Probabilistic Graphical Models

Probabilistic inference
Various inference tasks:

Diagnostic task. (from effect to cause)

P( Pneumonia | Fever = T )
Prediction task. (from cause to effect)

P( Fever | Pneumonia = T )
Other probabilistic queries (queries on joint distributions).
P (Fever )
P ( Fever , ChestPain )
CS 3710 Probabilistic Graphical Models

8
 Inference
Any query can be computed from the full joint distribution !!!
Joint over a subset of variables is obtained through
marginalization
P( A = a, C = c) = P( A = a, B = bi , C = c, D = d j )
i j

Conditional probability over set of variables, given other

variables values is obtained through marginalization and
definition of conditionals
P ( A = a , C = c, D = d )
P ( D = d | A = a, C = c ) =
P ( A = a, C = c )
P ( A = a , B = b , C = c, D = d )
i
= i

P( A = a, B = b , C = c, D = d )
i j
i j

CS 3710 Probabilistic Graphical Models

Inference.
Any query can be computed from the full joint distribution !!!
Any joint probability can be expressed as a product of
conditionals via the chain rule.

P( X 1 , X 2 ,K X n ) = P( X n | X 1, K X n 1 ) P( X 1, K X n 1 )
= P( X n | X 1, K X n 1 ) P( X n 1 | X 1, K X n 2 ) P( X 1, K X n 2 )

= i =1 P( X i | X 1, K X i 1 )
n

Sometimes it is easier to define the distribution in terms of

conditional probabilities:
E.g. P( Fever | Pneumonia = T )
P( Fever | Pneumonia = F )
CS 3710 Probabilistic Graphical Models

9
 Modeling uncertainty with probabilities
Defining the full joint distribution makes it possible to
represent and reason with uncertainty in a uniform way
We are able to handle an arbitrary inference problem
Problems:
Space complexity. To store a full joint distribution we
need to remember O(d n ) numbers.
n number of random variables, d number of values
Inference (time) complexity. To compute some queries
n
requires O(d. ) steps.
Acquisition problem. Who is going to define all of the
probability entries?

CS 3710 Probabilistic Graphical Models

Medical diagnosis example.

Space complexity.
Pneumonia (2 values: T,F), Fever (2: T,F), Cough (2: T,F),
WBCcount (3: high, normal, low), paleness (2: T,F)
Number of assignments: 2*2*2*3*2=48
We need to define at least 47 probabilities.
Time complexity.
Assume we need to compute the marginal of Pneumonia=T
from the full joint
P( Pneumonia = T ) =
= P( Fever = i, Cough = j, WBCcount = k , Pale = u )
iT , F jT , F k = h , n ,l uT , F

Sum over: 2*2*3*2=24 combinations

CS 3710 Probabilistic Graphical Models

10
 Graphical models
Aim: alleviate the representational and computational
bottlenecks
Idea: Take advantage of the structure, in particular,
independences and conditional independences that hold
among random variables

Two classes of models:

Bayesian belief networks
Modeling asymmetric (causal) effects and dependencies
Markov random fields
Modeling symmetric effects and dependencies among
random variables
Used often to model spatial dependences (image
analysis)
CS 3710 Probabilistic Graphical Models

Bayesian belief networks (BBNs)

Bayesian belief networks.
Represent the full joint distribution over the variables more
compactly with a smaller number of parameters.
Take advantage of conditional and marginal independences
among random variables

A and B are independent

P ( A, B ) = P ( A ) P ( B )
A and B are conditionally independent given C

P ( A, B | C ) = P ( A | C ) P ( B | C )
P( A | C , B) = P( A | C )

CS 3710 Probabilistic Graphical Models

11
 Alarm system example.
Assume your house has an alarm system against burglary.
You live in the seismically active area and the alarm system
can get occasionally set off by an earthquake. You have two
neighbors, Mary and John, who do not know each other. If
they hear the alarm they call you, but this is not guaranteed.
We want to represent the probability distribution of events:
Burglary, Earthquake, Alarm, Mary calls and John calls

Causal relations Burglary Earthquake

Alarm

JohnCalls MaryCalls

CS 3710 Probabilistic Graphical Models

Bayesian belief network.

1. Directed acyclic graph
Nodes = random variables
Burglary, Earthquake, Alarm, Mary calls and John calls
Links = direct (causal) dependencies between variables.
The chance of Alarm is influenced by Earthquake, The
chance of John calling is affected by the Alarm
Burglary Earthquake

Alarm

JohnCalls MaryCalls

CS 3710 Probabilistic Graphical Models

12
 Bayesian belief network.
2. Local conditional distributions
relate variables and their parents

Burglary P(B) Earthquake P(E)

Alarm P(A|B,E)

P(J|A) P(M|A)

JohnCalls MaryCalls

CS 3710 Probabilistic Graphical Models

Bayesian belief network.

P(B) P(E)
T F T F
Burglary 0.001 0.999 Earthquake 0.002 0.998

P(A|B,E)
B E T F
T T 0.95 0.05
Alarm T F 0.94 0.06
F T 0.29 0.71
F F 0.001 0.999
P(J|A) P(M|A)
A T F A T F
JohnCalls T 0.90 0.1 MaryCalls T 0.7 0.3
F 0.05 0.95 F 0.01 0.99

CS 3710 Probabilistic Graphical Models

13
 Bayesian belief networks (general)
Two components: B = ( S , S ) B E
Directed acyclic graph
Nodes correspond to random variables A
(Missing) links encode independences
J M
Parameters
Local conditional probability distributions
for every variable-parent configuration P(A|B,E)
B E T F
P ( X i | pa ( X i ))
T T 0.95 0.05
Where: T F 0.94 0.06
F T 0.29 0.71
pa ( X i ) - stand for parents of Xi F F 0.001 0.999

CS 3710 Probabilistic Graphical Models

Full joint distribution in BBNs

Full joint distribution is defined in terms of local conditional
distributions (obtained via the chain rule):

P ( X 1 , X 2 ,.., X n ) = P( X
i =1,.. n
i | pa ( X i ))

B E
Example:
Assume the following assignment A
of values to random variables
B = T, E = T, A = T, J = T, M = F J M

Then its probability is:

P( B = T , E = T , A = T , J = T , M = F ) =
P(B = T )P(E = T )P( A = T | B = T , E = T )P( J = T | A = T )P(M = F | A = T )
CS 3710 Probabilistic Graphical Models

14
 Bayesian belief networks (BBNs)
Bayesian belief networks
Represent the full joint distribution over the variables more
compactly using the product of local conditionals.
But how did we get to local parameterizations?
Answer:
Graphical structure encodes conditional and marginal
independences among random variables
A and B are independent P ( A , B ) = P ( A ) P ( B )
A and B are conditionally independent given C
P( A | C , B) = P( A | C )
P ( A, B | C ) = P ( A | C ) P ( B | C )
The graph structure implies the decomposition !!!
CS 3710 Probabilistic Graphical Models

Independences in BBNs
3 basic independence structures:

1. Burglary
2. 3.
Burglary Earthquake
Alarm

Alarm

Alarm JohnCalls MaryCalls

JohnCalls

CS 3710 Probabilistic Graphical Models

15
 Independences in BBNs

1. Burglary
2. 3.
Burglary Earthquake
Alarm

Alarm
Alarm JohnCalls MaryCalls

JohnCalls

1. JohnCalls is independent of Burglary given Alarm

P ( J | A, B ) = P ( J | A )
P ( J , B | A) = P ( J | A) P ( B | A)
CS 3710 Probabilistic Graphical Models

Independences in BBNs

1. Burglary
2. 3.
Burglary Earthquake
Alarm

Alarm

Alarm JohnCalls MaryCalls

JohnCalls

2. Burglary is independent of Earthquake (not knowing Alarm)

Burglary and Earthquake become dependent given Alarm !!
P(B, E ) = P(B)P(E )

CS 3710 Probabilistic Graphical Models

16
 Independences in BBNs

1. Burglary
2. 3.
Burglary Earthquake
Alarm

Alarm
Alarm JohnCalls MaryCalls

JohnCalls

3. MaryCalls is independent of JohnCalls given Alarm

P ( J | A, M ) = P ( J | A )
P ( J , M | A) = P ( J | A) P ( M | A)
CS 3710 Probabilistic Graphical Models

Independences in BBN
BBN distribution models many conditional independence
relations among distant variables and sets of variables
These are defined in terms of the graphical criterion called d-
separation
D-separation and independence
Let X,Y and Z be three sets of nodes
If X and Y are d-separated by Z, then X and Y are
conditionally independent given Z
D-separation :
A is d-separated from B given C if every undirected path
between them is blocked with C
Path blocking
3 cases that expand on three basic independence structures

CS 3710 Probabilistic Graphical Models

17
 Undirected path blocking
A is d-separated from B given C if every undirected path
between them is blocked

1. Path blocking with a linear substructure

Z
X Y

Z in C
X in A Y in B

CS 3710 Probabilistic Graphical Models

Undirected path blocking

A is d-separated from B given C if every undirected path
between them is blocked

2. Path blocking with the wedge substructure

Z

X Y
Z in C
X in A Y in B

CS 3710 Probabilistic Graphical Models

18
 Undirected path blocking
A is d-separated from B given C if every undirected path
between them is blocked

3. Path blocking with the vee substructure

X in A Y in B
Y
X
Z
Z or any of its descendants not in C

CS 3710 Probabilistic Graphical Models

Independences in BBNs
Burglary Earthquake

RadioReport
Alarm

JohnCalls MaryCalls

Earthquake and Burglary are independent given MaryCalls ?

CS 3710 Probabilistic Graphical Models

19
 Independences in BBNs
Burglary Earthquake

RadioReport
Alarm

JohnCalls MaryCalls

Earthquake and Burglary are independent given MaryCalls F

Burglary and MaryCalls are independent (not knowing Alarm) ?

CS 3710 Probabilistic Graphical Models

Independences in BBNs
Burglary Earthquake

RadioReport
Alarm

JohnCalls MaryCalls

Earthquake and Burglary are independent given MaryCalls F

Burglary and MaryCalls are independent (not knowing Alarm) F
Burglary and RadioReport are independent given Earthquake ?

CS 3710 Probabilistic Graphical Models

20
 Independences in BBNs
Burglary Earthquake

RadioReport
Alarm

JohnCalls MaryCalls

Earthquake and Burglary are independent given MaryCalls F

Burglary and MaryCalls are independent (not knowing Alarm) F
Burglary and RadioReport are independent given Earthquake T
Burglary and RadioReport are independent given MaryCalls ?

CS 3710 Probabilistic Graphical Models

Independences in BBNs
Burglary Earthquake

RadioReport
Alarm

JohnCalls MaryCalls

Earthquake and Burglary are independent given MaryCalls F

Burglary and MaryCalls are independent (not knowing Alarm) F
Burglary and RadioReport are independent given Earthquake T
Burglary and RadioReport are independent given MaryCalls F

CS 3710 Probabilistic Graphical Models

21
 Bayesian belief networks (BBNs)
Bayesian belief networks
Represents the full joint distribution over the variables more
compactly using the product of local conditionals.
So how did we get to local parameterizations?

P ( X 1 , X 2 ,.., X n ) = P( X
i =1,.. n
i | pa ( X i ))

The decomposition is implied by the set of independences

encoded in the belief network.

CS 3710 Probabilistic Graphical Models

Full joint distribution in BBNs

B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

CS 3710 Probabilistic Graphical Models

22
 Full joint distribution in BBNs
B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

= P(J = T | B = T, E = T, A = T, M = F)P(B = T, E = T, A = T, M = F)
= P(J = T | A = T)P(B = T, E = T, A = T, M = F)

CS 3710 Probabilistic Graphical Models

Full joint distribution in BBNs

B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

= P(J = T | B = T, E = T, A = T, M = F)P(B = T, E = T, A = T, M = F)
= P(J = T | A = T)P(B = T, E = T, A = T, M = F)
P(M = F | B = T, E = T, A = T)P(B = T, E = T, A = T)
P(M = F | A = T)P(B = T, E = T, A = T)

CS 3710 Probabilistic Graphical Models

23
 Full joint distribution in BBNs
B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

= P(J = T | B = T, E = T, A = T, M = F)P(B = T, E = T, A = T, M = F)
= P(J = T | A = T)P(B = T, E = T, A = T, M = F)
P(M = F | B = T, E = T, A = T)P(B = T, E = T, A = T)
P(M = F | A = T)P(B = T, E = T, A = T)
P( A = T | B = T, E = T)P(B = T, E = T)

CS 3710 Probabilistic Graphical Models

Full joint distribution in BBNs

B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

= P(J = T | B = T, E = T, A = T, M = F)P(B = T, E = T, A = T, M = F)
= P(J = T | A = T)P(B = T, E = T, A = T, M = F)
P(M = F | B = T, E = T, A = T)P(B = T, E = T, A = T)
P(M = F | A = T)P(B = T, E = T, A = T)
P( A = T | B = T, E = T)P(B = T, E = T)
P(B = T)P(E = T)

CS 3710 Probabilistic Graphical Models

24
 Full joint distribution in BBNs
B E
Rewrite the full joint probability using the
product rule: A

P(B = T, E = T, A = T, J = T, M = F) = J M

= P(J = T | B = T, E = T, A = T, M = F)P(B = T, E = T, A = T, M = F)
= P(J = T | A = T)P(B = T, E = T, A = T, M = F)
P(M = F | B = T, E = T, A = T)P(B = T, E = T, A = T)
P(M = F | A = T)P(B = T, E = T, A = T)
P( A = T | B = T, E = T)P(B = T, E = T)
P(B = T)P(E = T)
= P(J = T | A = T)P(M = F | A = T)P(A = T | B = T, E = T)P(B = T)P(E = T)
CS 3710 Probabilistic Graphical Models

25