 |
ARI711S - ARTIFICIAL INTELLIGENCE - 3RD OPP - JULY- AUG 2022 |
|
|
1 Page 1 |
▲back to top |
nAmlBIA ;unlVERSITY
OF SCIEnCEAnDTECHn OLOGY
FACULTYOF COMPUTING AND INFORMATICS
DEPARTMENT OF COMPUTER SCIENCE
QUALIFICATION: Bachelor of Computer Science
QUALIFICATION CODE: 07BACS
COURSE:Artificial Intelligence (and Computer Graphics)
SESSION:July 2022
DURATION: 3 Hours
LEVEL: 7
COURSECODE: ARl711S/AIG710S
PAPER:Theory
MARKS: 90
THIRD OPPORTUNITY EXAMINATION QUESTION PAPER
EXAMINER:
I Prof. Jose G. Quenum
MODERATOR:
I Mr Stantin Siebritz
This paper consists of 2 pages
(excluding this front page)
INSTRUCTIONS
1. This paper contains 4 questions.
2. Answer all questions on the exam paper.
3. Marks/scores are provided at the right end of each question
4. Do not use or bring into the examination venue books, mobile devices and other materials
that may provide you with unfair advantage. Should you be in possession of one right now,
draw the attention of the examination officer or the invigilator.
5. NUST examination rules and regulations apply.
PERMISSIBLEMATERIALS
Calculator
|
|
2 Page 2 |
▲back to top |
ARl711S
Third Exam (continued)
July 2022
Question 1 ..................................................................
[25 points]
(a) Consider an air cargo transport problem involving loading and unloading cargo and flying [15)
it from place to place. We use three actions in this problem: load, unload and fly. We
use two predicates to define the actions: in(x, y), which means that cargo x is inside
plane y; at(z, x), which means that object z (either cargo or plane) is at airport x. Note
that once inside a plane, a cargo is not considered at an airport any longer. Additionally,
the predicate cargo(x) means that x is a cargo; the predicate airport(y) means that y is
an airport and the predicate plane(z) means that z is a plane.
Initially we have three planes: Pi, P2 and P3 . We also have two cargos: C3 and C4 and
three airports: Loci and Loc4 and Loc5 . C3 is at Loci and C4 is at Loc4 . As well, Pi is at
Loci, P2 is at Loc4 and P3 is at Loe;;.
Using the STRIPS notation and first-order logic, define the actions and the initial knowl-
edge base.
(b) Consider the goal of moving C3 to Loc4 and C4 to Loci, update the partial plan
[10)
{unload(C 3 , Pi, Loc4 )} to satisfy the goal. Each step during the update must be discussed
and justified.
Question 2 ..................................................................
[20 points]
(a) The Millionaire is your favourite TV show. It is a ten-round game. Except for the first
[7]
round, the player can choose to play or quit at each round. When the player quits, the
game ends, and s/he can collect the rewards that s/he has earned so far. When the
player plays, s/he can succeed and move to the next round or fail, leading to the end of
the game. Note that ifs/he loses, all the rewards s/he has accumulated so far are lost.
Note also that when the player reaches the last round, whether s/he plays or not the
game ends with the appropriate reward.
Table 1: Millionaire - Rewards and success probability
Round Success Probability Reward
1
0.99
10
2
0.9
50
3
0.8
100
4
0.7
500
5
0.6
1000
6
0.5
5000
7
0.4
10000
8
0.3
50000
9
0.2
100000
10
0.1
500000
Model this problem as a Markov Decision process and evaluate the following policy: 1r =
{roundi H play, round 2 H play, round 3 H quit}. You will use a discount factor of
0.95.
Page 1 of 2
Please turn over to the next page ...
|
|
3 Page 3 |
▲back to top |
ARl711S
Third Exam (continued)
July 2022
(b) Applying the policy iteration algorithm, find the optimal policy for the problem starting
[13)
from the policy rr.
Question 3 ..................................................................
[25 points]
(a) A Sudoku puzzle of order 3 is a 9 x 9 grid filled with digits between 1 and 9. The following
[8]
rules apply. The same digit should not appear more than once in a column or a row. As
well, there should be no repetion of a digit in a 3 x 3 block.
Table 2: Incomplete Sudoku puzzle
26
81
3
7
8
6
4
5
7
5
1
7
9
39
51
4
3
2
5
1
3
2
5
2
4
9
38
46
Table represents an incomplete puzzle. Define the puzzle in Table as a constraint
satisfaction problem.
(b) Use forward checking and propagation to complete the puzzle. You will show all the
[17)
domain reductions.
Question 4 ..................................................................
[20 points]
The diagram in Figure represents an adversarial game. Using the a: - (3pruning, solve the
game. You will indicate the values of a: and (3at each node and where pruning occurs.
MIN
Page 2 of 2
4
Figure 1: Adversarial Search Problem
End of Exam