Artificial Intelligence

(2180703)

Unit-7 Statistical Reasoning

Rutal Mahajan
S.N.P.I.T.R.C., Umrakh
[email protected]

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 1
by Rutal Mahajan
Symbolic vs Statistical Reasoning
• Knowledge representation has to do with–
1. Representing knowledge in computers

2. Enable computers to “reason” with that knowledge.

• Two Types of Approaches

• Logical Approach:

Using common sense reasoning /Predicate logic /Propositional/ (any logical approach)

to represent facts, It assumes all values are either completely true or completely false.

• Statistical Approach:

Probabilistic approach (statistical) of knowledge representation considers

all values of sentence as

completely true 1, completely false 0 or (e.g. 0.5 –equal) chance of being true or

false

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 2
by Rutal Mahajan
Symbolic vs Statistical Reasoning

• Logical Approach
• E.g. Common sense reasoning:

* Knowledge: “Tweety is a bird”

Consequences: “Tweety flies”

In general, * (default rule) is applied. So according to that “tweety is bird so tweety can fly”

there is no evidence of abnormality of tweety can not fly.

- It can be represented using exceptions

- But “if the number of exceptions is large such system tend to break down.”

• Statistical Approach
- Provide a method for representing beliefs that are not certain (or uncertain) but for which

there may be some supporting (or contradictory) evidence. (works with Probabilities)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 3
by Rutal Mahajan
outline

Statistical Reasoning

Probability and Bayes’ Theorem

Certainty Factors and rule-based system

Bayesian Networks

Dampster-Shafer Theory

Fuzzy Logic

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 4
by Rutal Mahajan
Basic Statistical Method- Probability
Probability means 'likelihood' or 'chance‘ of
happening/ not happening of event.
Probability =
(number of desired outcomes) / (total number of outcomes)

The basic approach statistical methods adopt to deal with uncertainty is via the axioms
of probability

Probability axioms:
1. Probabilities are real numbers in range 0 to 1. P(A) є [0,1]
2. P(A)=0 indicates total uncertainty in A
P(A)=1 indicates total certainty in A and
values in between indicates some degree of (un)certainty.
3. Sum of all probabilities is 1. (total probability) P(U)=1

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 5
by Rutal Mahajan
Basic Statistical Method- Probability
Probability =
(number of desired outcomes) / (total number of outcomes)

Example:
A coin is tossed. What is the probability of getting a head?
Total number of outcomes of tossing coin (n): 2 (head and tail)
Number of outcome of our interest (m): 1 (we are interested in getting head)

P(head) =1/2

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 6
by Rutal Mahajan
Basic Statistical Method- Probability
Probability calculation of various types of events:
1. Mutually Exclusive Events:
If two events A and B are mutually exclusive (independent) of each other then,
probability of occurrence of either A or B can be given as:

P(A or B)= P(A) +P(B)

P(A U B) = P(A) +P(B) A B

Example: A card is drawn from a pack of 52, what is the probability that it is a king or a
queen?
Event (A): draw of king card Event (B): draw of queen card
P(card draw is king or queen) =P(A or B) = P( card is king) + P(card is queen)
= 4/52 + 4/52
= 2/13
2. Mutually Exclusive Events:
If two events A and B are mutually exclusive (independent) of each other then,
probability of either A or B can be given as:
Statistical Reasoning (AI B.E.8th semester)
3/28/2019 7
by Rutal Mahajan
Basic Statistical Method- Probability
Probability calculation of various types of events:
1. Mutually Exclusive Events:
If two events A and B are mutually exclusive (independent) of each other then, probability
of occurrence of either A or B can be given as:

P(A or B)= P(A) +P(B) A B

P(A U B) = P(A) +P(B)

Example: A card is drawn from a pack of 52, what is the probability that it is a king or a
queen?
Event (A): draw of king card Event (B): draw of queen card
P(card draw is king or queen) =P(A or B) = P( card is king) + P(card is queen)
= 4/52 + 4/52
= 2/13
2. Non- Mutually Exclusive Events: A and B
Cases of occurrences of event A or B and events are overlapping. A∩B
P(A or B)= P(A) +P(B)- P(A and B) A B
P(A U B) = P(A) +P(B) –P(A ∩ B)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 8
by Rutal Mahajan
Probability Theory- prerequisite
Basic probability Axioms are
P(U)=1
P(A) є [0,1]
P(A U B) = P(A) +P(B)-P(A ∩ B) (for mutually exclusive events P(A ∩ B) =0)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 9
by Rutal Mahajan
Probability Theory
• Single event – calculate simple probability

• Multiple events
1. Independent events (happening simultaneously)  calculate joint probability
This can be calculated by multiplying the probabilities of each event.
E.g. case of coin tossing twice,so probability of first and second toss both getting head
(½) * (½ ) =0.25
So in general, probability of two independent event A and B occurring together
P(A and B) =P(A) * P(B)

2. Dependent events  Conditional probability

Conditional or posterior probability of event A, given that event B has already occurred,
written as P(A|B)
P(A|B) = P(A ∩ B)/P(A)
P(A and B)
P(A) A B P(A ∩ B)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 10
by Rutal Mahajan
Probability Theory
Example:
A shooter is known to hit a target 3 out of 7 shots;
another shooter is known to hit the target 2 out of 5 shots.
Find the probability of the target being hit at all when both of them try.

Solution:

Probability of first shooter hitting the target P (A) = 3/7

Probability of second shooter hitting the target P (B) = 2/5

P(A and B)
P(A) A B P(A ∩ B)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 11
by Rutal Mahajan
Probability Theory
Example:
A shooter is known to hit a target 3 out of 7 shots;
another shooter is known to hit the target 2 out of 5 shots.
Find the probability of the target being hit at all when both of them try.

Solution:

Probability of first shooter hitting the target P (A) = 3/7

Probability of second shooter hitting the target P (B) = 2/5

Event A and B are not mutually exclusive as both the shooters may hit target.
Hence the additive rule applicable is

P(A∪B)=P(A)+P(B)−P(A∩B)
=3/7+2/5−((3/7)×(2/5))=29/35−6/35=23/35

P(A and B)
P(A) A B P(A ∩ B)

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 12
by Rutal Mahajan
Probability Theory- Conditional Probability
Example:
In a school there are 500 students.
400 students are tall, others are short.
300 students are white, others are colored. 200 students are both tall and white.
Find probability of student being tall if he is white.

Solution:
T: the student is tall
W: the student is white
P(T|W) = ?

white 200 200

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 13
by Rutal Mahajan
Probability Theory- Conditional Probability
Example:
In a school there are 500 students.
400 students are tall, others are short.
300 students are white, others are colored. 200 students are both tall and white.
Find probability of student being tall if he is white.

Solution:
T: the student is tall
W: the student is white
P(T|W) = ?

Probability of student being tall if he is white

P(T|W) =P(T ∩ W)/ P(W)
=200/300
=0.66

white 200 200

Tall

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 14
by Rutal Mahajan
Probability & Bayes’ Theorem
P(Hi|E) = the probability that hypothesis Hi is true given evidence E

P(E | Hi) = the probability that we will observe evidence E given that hypothesis Hi is
true

P(Hi) = the a priori probability that hypothesis Hi is true in the absence of any specific
evidence. These probabilities are called prior probabilities or priories.

k= the number of possible hypothesis

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 15
by Rutal Mahajan
Bayesian Belief Network- BBN
Refere BBN ppt.

Statistical Reasoning (AI B.E.8th semester)

3/28/2019 16
by Rutal Mahajan
Other questions from GTU

1. Enlist and explain different approaches to deal with uncertainty.

2. Explain conditional probability and its use in AI.
or
Write a short note on Conditional probability
3. Write a short note on joint probability
4. What is Bayes theorem? Explain with example.
5. Explain the use of probabilistic reasoning in medical diagnosis
6. Define certainty factor. How does the certainty factor helps in dealing with the
uncertainty?
Or
Write a short note on certainty factor and rule based system
7. What are belief networks? Explain steps to build belief network with an example
8. Explain Dempster-Shafer theory to deal with uncertainty.
9. Write a short note on Fuzzy logic

Using Predicate Logic (AI B.E.8th semester)

3/28/2019 17
by Rutal Mahajan
 References

1. Artificial Intelligence, 3rd Edition, Elaine Rich, Kevin Knight,

Shivshankar B Nair
2. NPTEL Notes and video lectures

Using Predicate Logic (AI B.E.8th semester)

3/28/2019 18
by Rutal Mahajan