CS-171, Intro to A.I.

— Mid-term Exam — Fall Quarter, 2013

YOUR NAME AND ID NUMBER:

YOUR ID: ID TO RIGHT: ROW: NO. FROM RIGHT:

The exam will begin on the next page. Please, do not turn the page until told.

When you are told to begin the exam, please check first to make sure that you
have all seven pages, as numbered 1-7 in the bottom-right corner of each page.

The exam is closed-notes, closed-book. No calculators, cell phones, electronics.

Please clear your desk entirely, except for pen, pencil, eraser, a blank piece of
paper (for scratch pad use), and an optional water bottle. Please write your name
and ID# on the blank piece of paper and turn it in with your exam.

This page summarizes the points available for each question so you can plan your time.

1. (10 pts total, -2 pts each error, but not negative) STATE SPACE SEARCH.

2. (5 pts total, -2 pts for each error, but not negative) TASK ENVIRONMENT.

3. (5 pts total, -1 pt each wrong answer, but not negative) SEARCH PROPERTIES.

4. (15 pts total, 5 pts each) ALPHA-BETA PRUNING NODE ORDERING.

5. (20 pts total, -5 for each error, but not negative) RESOLUTION THEOREM
PROVING.

6. (20 points total, 4 pts each) CONSTRAINT SATISFACTION PROBLEMS.

7. (25 pts total, 5 pts each) SEARCH.

1
2
 1. (10 pts total, -2 pts each error, but not negative) STATE SPACE SEARCH. The
Missionaries and Cannibals problem is a classic brainteaser. It is a member of a class
of river-crossing brainteasers that includes Farmer to Market, Jealous Husbands, and
others (see https://en.wikipedia.org/wiki/River-crossing_problems). See Section 3.1.1 and Figure 3.3

Three missionaries and three cannibals must cross a river using a boat which
can carry at most two people. The constraint is that, for both river banks, if there are
missionaries present on the bank, they cannot be outnumbered by cannibals (if they
were, the cannibals would eat the missionaries). The boat cannot cross the river by
itself without people on board. How can all six people get safely across the river?

We will call states where anyone gets eaten “forbidden” and consider them as part of
the state space, but from which there is no return; i.e., there is no way to go to any other
state (including back to the previous state) from such a forbidden state. All other states
are “allowed.” It is always possible to go from an allowed state to some other state(s).
One way to represent this problem is as a state vector with components (M,C,B),
where M=the number of missionaries on the wrong side, C=the number of cannibals on
the wrong side, and B=the number of boats on the wrong side. Since everything starts
on the wrong side, the start state is (3,3,1). The single goal state is (0,0,0).

1.a. (2 pts) What is the maximum branching factor counting BOTH allowed and
forbidden states as children?
The boat can carry (1) a missionary alone, (2) a
cannibal alone, (3) a missionary and a cannibal,
5 .
(4) two missionaries, or (5) two cannibals.
1.b. (8 pts, -2 for each error, but not negative) Draw the state space showing You
ONLY are the
not obliged to
ALLOWED states. I.e., do NOT show the FORBIDDEN states. draw the state space
in a straight line. If
Represent an allowed state as a circle enclosing its state vector (M,C,B).
Connect allowed states that are possible successors of each other. The start your
and graph
goal is correct
topologically, you will
states are shown. The first node expansion is done for you as an example.
receive full credit.

2,2,0 You see that a good representation can make 1,1,1

this brainteaser become almost trivial to solve.

3,3,1 3,1,0 3,2,1 3,0,0 3,1,1 1,1,0 2,2,1 0,2,0 0,3,1 0,1,0 0,2,1 0,0,0

3,2,0 This node can be reached only from the goal node, when a cannibal then
0,1,1
crosses the river again alone. It would never be reached in search because
search would stop at the goal node. You are not obliged to draw it.

If you solved the problem

**** TURN PAGE“in English,”
OVER ANDi.e.,CONTINUE
wrote out theON
boatTHE
moves in English
OTHER that****
SIDE solved the
brainteaser, then you will receive some credit, but very little. The point of the problem was to
reason about transitions in state space, and to illustrate the general and powerful state-space
approach to problem-solving --- not to get the missionaries and cannibals safely across the river. 3
2. (5 pts total, -2 pts for each error, but not negative) TASK ENVIRONMENT. Your
book defines a task environment as a set of four things, with the acronym PEAS. Fill in
the blanks with the names of the PEAS components.
See Section 2.3.1.
Performance (measure) Environment Actuators Sensors

3. (5 pts total, -1 pt each wrong answer, but not negative) SEARCH PROPERTIES.
Fill in the values of the four evaluation criteria for each search strategy shown. Assume
a tree search where b is the finite branching factor; d is the depth to the shallowest goal
node; m is the maximum depth of the search tree; C* is the cost of the optimal solution;
step costs are identical and equal to some positive ε; and in Bidirectional search both
directions using breadth-first search. See Figure 3.21.
Note that these conditions satisfy all of the footnotes of Fig. 3.21 in your book.
Criterion Complete? Time complexity Space complexity Optimal?
Breadth-First Yes O(b^d) O(b^d) Yes
Uniform-Cost Yes O(b^(1+floor(C*/ε))) O(b^(1+floor(C*/ε))) Yes
O(b^(d+1)) also OK O(b^(d+1)) also OK
Depth-First No O(b^m) O(bm) No
Iterative Deepening Yes O(b^d) O(bd) Yes
Bidirectional Yes O(b^(d/2)) O(b^(d/2)) Yes
(if applicable)

4. (15 pts total, 5 pts each) ALPHA-BETA PRUNING NODE ORDERING. Alpha-beta
pruning effectiveness depends on the order in which node values are encountered. A
bad node order can yield no pruning. A good node order can yield maximal pruning.
This question asks about node ordering in relation to alpha-beta pruning effectiveness.
In the game tree below, the agent has searched the left branch and determined it
is worth 5. It is about to search the right branch. How does the left-to-right order of the
node values on the right branch affect how many nodes can be pruned in that branch?
Below, write A, B, and C in an order that gives a correct answer to each question.

(Max) α=5
β = +oo
See Section 5.3. You are only obliged
α=5 to provide one
(Min) 5
β = +oo correct node order
for each question.
(A) (B) (C) However, for each
7 9 5 8 4 6 question there are
two correct answers.

4.a. (5 pts) What node ordering prunes two nodes? B, A, C or B, C, A

4.b. (5 pts) What node ordering prunes one node? A, B, C or C, B, A

4.c. (5 pts) What node ordering prunes zero nodes? A, C, B or C, A, B

4
5. (20 pts total, -5 for each error, but not negative) RESOLUTION THEOREM PROVING.
You are engaged in Knowledge Engineering for the Wumpus Cave. You have interviewed an
expert on the Wumpus Cave who told you, among other things, “A breeze in square (1,1) is
equivalent to a pit in square (1,2) or a pit in square (2,1).” You translated this into propositional
logic as, “(B11 ⇔ P12 ∨ P21),” and then into Conjunctive Normal Form as “(¬B11 ∨ P12 ∨ P21)
∧ (¬P12 ∨ B11) ∧ (¬P21 ∨ B11).”
Now it is time for the first “live” test of your system. An agent has been lowered down
into the Wumpus cave, and reports back by radio, “Square (1,1) has a breeze. Also, I went into
square (1,2) and I did not die, so it does not have a pit.” You translate this knowledge into
propositional logic as “(B11) ∧ (¬ P12)” and add it to your knowledge base.
Next your system is asked to perform inference. The agent asks by radio, “Is it true that
square (2,1) has a pit?” You translate this query into propositional logic as the goal sentence
“(P21).” You form the negated goal as “(¬ P21).” Your knowledge base plus negated goal is:
(¬B11 ∨ P12 ∨ P21)
(¬P12 ∨ B11)
(¬P21 ∨ B11) See Sections 7.2 and 7.5.2.
(B11)
(¬ P12)
(¬ P21)
Run resolution on this knowledge base until you produce the null clause, “( )”,
thereby proving that the goal sentence is true. The shortest proof I know of is only three
lines long. It is OK to use more lines, if your proof is correct. SHOW YOUR WORK.
Repeatedly choose two clauses, write one clause in the first blank space on a line,
and the other clause in the second. Apply resolution to them. Write the resulting clause
in the third blank space, and insert it into the knowledge base.

Resolve (¬B11 ∨ P12 ∨ P21) and (B11) to give (P12 ∨ P21)

Resolve (P12 ∨ P21) and (¬P12) to give (P21)

Resolve (P21) and (¬P21) to give ()

Resolve and to give

Other proofs are OK as long as they are correct.

Resolve Forand
example, you might performto the resolution
give
steps above in any other order you choose.

Resolve and to give

**** TURN PAGE OVER AND CONTINUE ON THE OTHER SIDE ****

5
6. (20 points total, 4 pts each) CONSTRAINT SATISFACTION PROBLEMS.

See Chapter 6.
TN NC

AL GA

FL SC

You are a map-coloring robot assigned to color this Southeast USA map. Adjacent regions must
be colored a different color (R=Red, B=Blue, G=Green). The constraint graph is shown.

6a. (4pts total, -2 each wrong answer, but not negative) FORWARD CHECKING.
Cross out all values that would be eliminated by Forward Checking, after variable GA
has just been assigned value G, as shown:
AL TN FL GA NC SC
X
RGB X
RGB RGBX G RGB X RGB X
6b. (4pts total, -2 each wrong answer, but not negative) ARC CONSISTENCY.
AL and FL have been assigned values, but no constraint propagation has been done.
Cross out all values that would be eliminated by Arc Consistency (AC-3 in your book).
AL TN FL GA NC SC
B XX
RGB R X X
RGB XX
RGB RGB XX
6c. (4pts total, -2 each wrong answer, but not negative) MINIMUM-REMAINING-
VALUES HEURISTIC. Consider the assignment below. TN is assigned and constraint
propagation has been done. List all unassigned variables that might be selected by the
Minimum-Remaining-Values (MRV) Heuristic: AL, GA, NC .

AL TN FL GA NC SC
RB G RGB RB RB RGB

6d. (4pts total, -2 each wrong answer, but not negative) DEGREE HEURISTIC.
Consider the assignment below. (It is the same assignment as in problem 6c above.)
TN is assigned and constraint propagation has been done. List all unassigned variables
that might be selected by the Degree Heuristic:. GA .

AL TN FL GA NC SC
RB G RGB RB RB RGB

6e. (4pts total) MIN-CONFLICTS HEURISTIC. Consider the complete but inconsistent
assignment below. GA has just been selected to be assigned a new value during local
search for a complete and consistent assignment. What new value would be chosen
below for GA by the Min-Conflicts Heuristic?. R .

AL TN FL GA NC SC
B G R ? G B

6
 7 (25 pts total, 5 pts each) SEARCH. Execute Tree Search through this graph (i.e., do not remember
visited nodes, so repeated nodes are possible). It is not a tree, but pretend that you don’t know that.
Step costs are given next to each arc. Heuristic values are given next to each node (as h=x). The
successors of each node are indicated by the arrows out of that node.
Successors are returned in left-to-right order. (Note: B is a successor of itself).
For each search strategy below, show the orderPlease
in which
seenodes are expanded
the lecture (i.e.,
slides for to expand aSearch, topic
Uninformed
node means that its children are generated), ending with“When
the goal node that is found. Show
to do Goal-Test? When generated? the path from
When popped?”
start to goal, or write “None”. The first one is done for for
youclarification
as an example.about exactly what to do in practical cases.
S h=20
5 100
See Chapter 3. 20 B
h=25 14 h=21
A

30 26
G1 G2
7a. DEPTH FIRST SEARCH.
DFS does the Goal-test before the
See Section 3.4.3 child is pushed onto the queue. The
Order of node expansion: S A G1
and Fig. 3.17. goal is found when A is expanded.
Path found: S A G1
7b. (5 pts)BREADTH FIRST SEARCH.
BFS does the Goal-test before the
See Section 3.4.1
Order of node expansion: S A G1 child is pushed onto the queue. The
and Fig. 3.11.
goal is found when A is expanded.
Path found: S A G1
7c. (5 pts) UNIFORM COST SEARCH.
See Section 3.4.2 UCS does goaltest when node is popped off queue.
and Fig. 3.14.Order of node expansion: S B A A G2
Path found: S B A G2
7d. (5 pts) GREEDY (BEST-FIRST) SEARCH.
See Section 3.5.1
and Fig. 3.23.Order of node expansion: S B B B B B B ... etc. B always has lower h(=21) than any other node on queue.

Path found: None

7d. (5 pts) ITERATED DEEPENING SEARCH.
See Sections 3.4.4-5 IDS does goaltest when node is generated.
Order of node expansion: S S A G1 Goal G1 is found when A is expanded.
and Figs. 3.18-19.
Path found: S A G1
7e. (5 pts) A* SEARCH.

See Section 3.5.2

Order of node expansion: S B A A G2 A* does goaltest when node is popped off queue.
and Figs. 3.24-25.
Path found: S B A G2

**** THIS IS THE END OF THE MID-TERM EXAM ****

7