CS 540: Introduction to Artificial Intelligence

Final Exam: 2:45-4:45pm, May 11, 2008

Room 1240 Computer Sciences

CLOSED BOOK
(two sheets of notes and a calculator allowed)

Write your answers on these pages and show your work. If you feel that a question is not fully specified,
state any assumptions that you need to make in order to solve the problem. You may use the backs of
these sheets for scratch work. There also is one blank page at the end of the exam.

Write your name on this and all other pages of this exam. Make sure your exam contains seven problems
on eleven pages.

Name ________________________________________________________________

Student ID ________________________________________________________________

Problem Score Max Score

1 ______ 18

2 ______ 15

3 ______ 25

4 ______ 10

5 ______ 5

6 ______ 12

7 ______ 15

TOTAL ______ 100

 Name: _____________________________

Problem 1 – Reasoning with Probability Distributions (18 points)

a) Consider the joint probability distribution below.

A B C P(A, B, C)
False False False 0.10
False False True 0.13
False True False 0.15
False True True 0.12
True False False 0.01
True False True 0.19
True True False 0.07
True True True 0.23

i. What is P(A = false)? ______________ Show your work below.

ii. What is P(A=true or B=true)? ______________ Explain.

2
 Name: _____________________________

b) Consider the task of diagnosing a human disease D, given two symptoms, A and B, where we
have gathered some historical data from a large population of people. (A, B and D are all
Boolean-valued.)
We found that when D=true, then A=true 60% of the time.
Also, when D=true, we found that B=true 20% of the time.
When D=false, then A=true 30% of the time.
When D=false, then B=true 80% of the time.
Finally, in this data D is true 10% of the time.
Using this information, estimate whether D is more likely true or false, for a new person
where A=true and B=false. Show and justify your calculations, including any assumptions
you need to make.

3
 Name: _____________________________

Problem 2 – Bayesian Networks (15 points)

Consider the following Bayesian Network, where variables A-D are all Boolean-valued:

A P(A=true) = 0.4 B P(B=true) = 0.6

C
A B P(C =true | A, B)
false false 0.8 D
false true 0.3
true false 0.4
true true 0.7
C P(D=true | C)
false 0.9
true 0.7

a) What is the probability that all four of these Boolean variables are true? ______________
[Be sure to show your work for Parts a-c.]

b) What is the probability that A is false and D is false? _______________

c) What is the probability that B is true given that C is false and D is true? _______________

4
 Name: _____________________________

Problem 3 – Representing Knowledge with First-Order Logic (25 points)

a) Convert each of the following English sentences into First-Order Predicate Calculus
(FOPC), using reasonably named predicates, functions, and constants. If you feel a sentence
is ambiguous, clarify which meaning you’re representing in logic. (Write your answers in
the space below each English sentence.)

Some person is in both CS 538 and CS 540 during Spring term 2008.

Every dog hates the same cat.

Driving to work changes one’s location but not one’s boss

[You must use situation calculus here.]

5
 Name: _____________________________

b) Provide a formal interpretation that shows that the following translation from English to
FOPC is incorrect. Be sure to explain your answer formally using the interpretation you
provide.

There exists a blue dress. x [ dress(x) blue(x) ]

c) What is the most-general unifier (mgu) of these two wff’s? ____________________

P(f(?x), ?y, g(?y)) P(?a, h(1, 2, ?b), g(?c))

Justify that your mgu unifies these two expressions:

6
 Name: _____________________________

Problem 4 – Logical Reasoning (10 points)

Given the following background knowledge:

1. x { [ p(x) q(x) ]  [ r(x) s(x) ] } ______________________________________

2. y [ q(y)  s(y) ] ______________________________________

3. p(1) ______________________________________

4. p(2) ______________________________________

5. q(2) ______________________________________

Show: z [ r(z) ]

Do this by putting wff’s 1-5 into CNF and use the resolution inference rule. Put the CNF on the
lines above, to the right of the wff’s. Put the CNF that results from the query below.

6. ______________________________________

7
 Name: _____________________________

Problem 5 – Neural Networks (5 points)

Consider a perceptron that has three real-valued inputs and an output unit with a step function as
its activation function. All the initial weights and the bias (“threshold”) equal -1.0. Assume the
teacher has said that the output should be 1 for the input in1 = -2, in2 = 0, and in3 = 4.

Show how the perceptron learning rule would alter this neural network upon processing this
training example. Let (the learning rate) be 0.2 and be sure to adjust the output unit’s bias
during training.

Perceptron BEFORE Training

Perceptron AFTER Training

8
 Name: _____________________________

Problem 6 – Important AI Concepts (12 points)

Describe each of the following AI concepts and briefly explain its most significant aspect.
(Write your answers in the space below the AI concept.)

Similarity Functions
description:

significance:

CLOSED Lists
description:

significance:

Semi-Decidability
description:

significance:

Alpha-Beta Pruning
description:

significance:

9
 Name: _____________________________

Problem 7– Learning What a Neural Network Learned (15 points)

Assume we have successfully trained the simple neural network below. The output unit uses a
“step function,” whose “threshold” is 7, to compute its output given its weighted input. The
inputs are binary-valued (i.e., either 0 or 1).
.

X1
9 =7
4
X2 Y

6
X3

a) We would like to understand what this neural network has learned. So we decide to
address this task by giving all of the eight possible input vectors to this neural network,
collecting this set of input/output pairs into a training set. Complete the contents of this
training set below:

X1 X2 X3 Y

0 0 0 ___
0 0 1 ___
0 1 0 ___
0 1 1 ___
1 0 0 ___
1 0 1 ___
1 1 0 ___
1 1 1 ___

[Feel free to use the empty space around this

training set for your calculations for Part b.]

10
 Name: _____________________________

b) We next decide to learn a decision tree that fully fits Part a’s training set, since decision
trees are often easy to interpret. Show all the calculations needed to create this tree, and
show the final decision tree that results. (Feel free to do some of your calculations on the
previous page.)

c) Should we worry about pruning Part b’s decision tree? Why or why not?

Have a good summer!

11