Uncertainty

CHAPTER 13
 Outline

Uncertainty
Probability
Syntax and Semantics
Inference
Independence and Bayes' Rule
 Uncertainty

Let action At = leave for airport t minutes before flight

Will At get me there on time?
1. partial observability (road state, other drivers' plans, etc.)
2. uncertainty in action outcomes (flat tire, etc.)
Hence a purely logical approach either
1. risks falsehood: “A25 will get me there on time”, or
2. leads to conclusions that are too weak for decision making:

“A25 will get me there on time if there's no accident on the bridge and it doesn't rain
and my tires remain intact etc etc.”
(A1440 might reasonably be said to get me there on time but
I'd have to stay overnight in the airport …)

1. immense complexity of modeling and predicting traffic

 noisy
sensors (traffic reports)
Problems:
 Methods for handling uncertainty
 Default or nonmonotonic logic:
 Assume my car does not have a flat tire
 Issues: What assumptions are reasonable? How to handle
contradiction?
 Rules with fudge factors:
 A25 |→0.3 get there on time
 Sprinkler |→ 0.99 WetGrass
 Issues: Problems with combination, e.g., Sprinkler causes Rain??
 Given the available evidence,

 A25 will get me there on time with probability 0.04

• Probability
• Model agent's degree of belief
 WetGrass |→ 0.7 Rain
 Assume A25 works unless contradicted by evidence
 Probability

Probabilistic assertions summarize effects of

 laziness: failure to enumerate exceptions, qualifications, etc.

Subjective probability:
 Probabilities relate propositions to agent's own state of
knowledge
e.g., P(A25 | no reported accidents) = 0.06

These are not assertions about the world

e.g., P(A25 | no reported accidents, 5 a.m.) = 0.15
• Probabilities of propositions change with new evidence:
 ignorance: lack of relevant facts, initial conditions,
etc.
 Making decisions under uncertainty

Suppose I believe the following:

P(A25 gets me there on time | …) = 0.04
P(A90 gets me there on time | …) = 0.70
P(A120 gets me there on time | …) = 0.95
P(A1440 gets me there on time | …) = 0.9999
 Which action to choose?
• Depends on my preferences for missing flight vs. time
spent waiting, etc.
• Utility theory is used to represent and infer
preferences
• Decision theory = probability theory + utility theory

 Syntax

 Basic element: random variable

 Similar to propositional logic: possible worlds defined by
assignment of values to random variables.
 Boolean random variables e.g., Cavity (do I have a cavity?)
 Domain values must be exhaustive and mutually exclusive
 Elementary proposition constructed by assignment of a
value to a random variable: e.g., Weather = sunny, Cavity
= false (abbreviated as cavity)
 Complex propositions formed from elementary
propositions and standard logical connectives e.g., Weather
= sunny  Cavity = false
Discrete random variables e.g., Weather is one of
<sunny,rainy,cloudy,snow>
 Syntax

Atomic event: A complete specification of the state of

the world about which the agent is uncertain
E.g., if the world consists of only two Boolean variables Cavity
and Toothache, then there are 4 distinct atomic events:
Cavity = false Toothache = false
Cavity = false  Toothache = true
Cavity = true  Toothache = false
Cavity = true  Toothache = true

Atomic events are mutually exclusive and exhaustive

•


 Axioms of probability

For any propositions A, B

 0 ≤ P(A) ≤ 1
 P(true) = 1 and P(false) = 0
 P(A  B) = P(A) + P(B) - P(A  B)

 Prior probability
 Prior or unconditional probabilities of propositions

 Joint probability distribution for a set of random variables gives the probability of every atomic
event on those random variables

Weather = sunny rainy cloudy snow

Cavity = true 0.144 0.02 0.016 0.02
Cavity = false0.576 0.08 0.064 0.08
 Every question about a domain can be answered by the joint distribution
•

• P(Weather,Cavity) = a 4 × 2 matrix of values:

•
e.g., P(Cavity = true) = 0.1 and P(Weather = sunny) = 0.72
correspond to belief prior to arrival of any (new) evidence
Probability distribution gives values for all possible assignments:
P(Weather) = <0.72,0.1,0.08,0.1> (normalized, i.e., sums to 1)
 Conditional probability

 Conditional or posterior probabilities

e.g., P(cavity | toothache) = 0.8
 (Notation for conditional distributions:
P(Cavity | Toothache) = 2-element vector of 2-element vectors)
 If we know more, e.g., cavity is also given, then we have
P(cavity | toothache,cavity) = 1
 New evidence may be irrelevant, allowing simplification, e.g.,
P(cavity | toothache, sunny) = P(cavity | toothache) = 0.8

 This kind of inference, sanctioned by domain knowledge, is crucial


 i.e., given that toothache is all I know

 Conditional probability

 Definition of conditional probability:

 Product rule gives an alternative formulation:
 A general version holds for whole distributions, e.g.,
 (View as a set of 4 × 2 equations, not matrix mult.)
 Chain rule is derived by successive application of product rule:
= P(X1,...,Xn-2) P(Xn-1 | X1,...,Xn-2) P(Xn | X1,...,Xn-1)
=…
= πi= 1^n P(Xi | X1, … ,Xi-1)

P(X1, …,Xn) = P(X1,...,Xn-1) P(Xn | X1,...,Xn-1)

P(Weather,Cavity) = P(Weather | Cavity) P(Cavity)
P(a  b) = P(a | b) P(b) = P(b | a) P(a)
P(a | b) = P(a  b) / P(b) if P(b) > 0
 Inference by enumeration

 Start with the joint probability distribution:

 For any proposition φ, sum the atomic events where it is

true: P(φ) = Σω:ω╞φ P(ω)



 Inference by enumeration

 Start with the joint probability distribution:

 For any proposition φ, sum the atomic events where it is

true: P(φ) = Σω:ω╞φ P(ω)
 P(toothache) = 0.108 + 0.012 + 0.016 + 0.064 = 0.2



 Inference by enumeration

 Start with the joint probability distribution:

 For any proposition φ, sum the atomic events where it is

true: P(φ) = Σω:ω╞φ P(ω)
 P(toothache) = 0.108 + 0.012 + 0.016 + 0.064 = 0.2



 Inference by enumeration

 Start with the joint probability distribution:

 Can also compute conditional probabilities:

P(cavity | toothache) = P(cavity  toothache)
P(toothache)
= 0.016+0.064
0.108 + 0.012 + 0.016 + 0.064
= 0.4



 Normalization

 Denominator can be viewed as a normalization constant α

P(Cavity | toothache) = α, P(Cavity,toothache)

= α, [P(Cavity,toothache,catch) + P(Cavity,toothache, catch)]
= α, [<0.108,0.016> + <0.012,0.064>]
= α, <0.12,0.08> = <0.6,0.4>

General idea: compute distribution on query variable by fixing evidence

variables and summing over hidden variables


 Inference by enumeration, contd.

Typically, we are interested in the posterior joint distribution of the query

variables Y
given specific values e for the evidence variables E
 The terms in the summation are joint entries because Y, E and H
together exhaust the set of random variables
1. Worst-case time complexity O(dn) where d is the largest arity
 Space complexity O(dn) to store the joint
distribution
 How to find the numbers for O(dn) entries?
Obvious problems:
Let the hidden variables be H = X - Y - E
Then the required summation of joint entries is done by
summing out the hidden variables:
P(Y | E = e) = αP(Y,E = e) = αΣhP(Y,E= e, H = h)
 Independence

 A and B are independent iff

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

P(Toothache, Catch, Cavity, Weather)

= P(Toothache, Catch, Cavity) P(Weather)

 32 entries reduced to 12; for n independent biased coins, O(2n) →O(n)

• Absolute independence powerful but rare
• Dentistry is a large field with hundreds of variables, none
of which are independent. What to do?

 Conditional independence

 P(Toothache, Cavity, Catch) has 23 – 1 = 7 independent entries

 If I have a cavity, the probability that the probe catches in it doesn't
depend on whether I have a toothache:
(1) P(catch | toothache, cavity) = P(catch | cavity)

 The same independence holds if I haven't got a cavity:

(2) P(catch | toothache,cavity) = P(catch | cavity)
 Catch is conditionally independent of Toothache given Cavity:
P(Catch | Toothache,Cavity) = P(Catch | Cavity)

 Equivalent statements:
P(Toothache | Catch, Cavity) = P(Toothache | Cavity)
P(Toothache, Catch | Cavity) = P(Toothache | Cavity) P(Catch | Cavity)


 Conditional independence contd.

 Write out full joint distribution using chain rule:

P(Toothache, Catch, Cavity)
= P(Toothache | Catch, Cavity) P(Catch, Cavity)
I.e., 2 + 2 + 1 = 5 independent numbers
 In most cases, the use of conditional independence reduces
the size of the representation of the joint distribution from
exponential in n to linear in n.
Conditional independence is our most basic and robust
form of knowledge about uncertain environments.
•
 = P(Toothache | Catch, Cavity) P(Catch | Cavity)
P(Cavity)
 = P(Toothache | Cavity) P(Catch | Cavity) P(Cavity)

 Bayes' Rule

 Product rule P(ab) = P(a | b) P(b) = P(b | a) P(a)

 Bayes' rule: P(a | b) = P(b | a) P(a) / P(b)
 or in distribution form

 Useful for assessing diagnostic probability from causal

probability:
 P(Cause|Effect) = P(Effect|Cause) P(Cause) / P(Effect)
 E.g., let M be meningitis, S be stiff neck:
P(m|s) = P(s|m) P(m) / P(s) = 0.8 × 0.0001 / 0.1 = 0.0008
 Note: posterior probability of meningitis still very small!

•
P(Y|X) = P(X|Y) P(Y) / P(X) = αP(X|Y) P(Y)

 Bayes' Rule and conditional independence

P(Cavity | toothache  catch)

= αP(toothache  catch | Cavity) P(Cavity)
= αP(toothache | Cavity) P(catch | Cavity) P(Cavity)

 This is an example of a naïve Bayes model:

 Total number of parameters is linear in n

• P(Cause,Effect1, … ,Effectn) = P(Cause) πiP(Effecti|
Cause)
 Summary

Probability is a rigorous formalism for uncertain

knowledge
Queries can be answered by summing over atomic
events
For nontrivial domains, we must find a way to
reduce the joint size
Independence and conditional independence
provide the tools
Joint probability distribution specifies probability of
every atomic event
Use forward chaining and backward chaining to
solve for the given knowledge base and goal.
Knowledge Base
 If [X croaks and eats flies] Then [X is a Toad]
 If [X chirps and sings] Then [X is a canary]
 If [X is a toad] Then [X is colored brown]
 If [X is a canary] Then [X is colored yellow]
 [John croaks and eats flies]
Goal: [John is colored Y]?