Artificial Intelligence - Test I

Fcbruary 6, 2024

Answer all the questions.

Total marks: 50
1. The Missionary and Cannibals (or
Hobbits and Orcs) problem is defined as follows: At one side
of a river there are three missionaries and
three cannibals. They have a boat that can transport
at most two persons. The goal for all persons is to cross
the river. At no tiune should the number
of cannibals exceed the number of
missionaries for obvious reasons. (15 Marks]

Missionaries

Cannibals

(a) Define the state space, action space, goal states and action cost
(b) Solve the problem via DFS and BFS by annotating the graph with numbers.
(c) Consider a heuristic function that counts the number of people on the other side of the
river. Is this heuristic consistent ? Is this heuristic admissible ? If so run the A search.
2. Define consisteut heuristics and admissible heuristics. Give an example of an admissible heuristic
which is not consistent [10 Marks
3. Prove the following statements [10 Marks]
(a) The maximum of two admissible heuristics is an admissible heuristic.
(b) The maximum of two consistent heuristics is a consistent heuristic.
4. What is the running time of uniform cost search (in terms of d, m, b and C"). Explain. Is
uniform cost search OPTIMAL ? (specify the conditions required) (5 Marks]
5. Consider the following search algorithm [10Marks)
(a) Set d+-1
(b) WHILE d<E m
Run DFS(d). The DFS(d) will expand up to depth d. During this process, DFS will
check IF goal state is visited. If so return the path cost and exit

1
(d) ELSE Set d -- d+1 and continue with the while loop
Answer the following: (Justify each cAse)
(a) Is the algorithrm OPTIMAL? (specify the conditions required)
(b) ls the algoritm COMPLETE?
(c) What is the time complexity (in terms of m, d and b) ?
(d) What is the space conplexity (in terms of m, dand b)?
(e) Is this approach better than BFS (Breadth First Search) ?
 CSSTL Artifical Intelligenee

March 19, 2024

Total Marks: 40 Marks

1. Prove the following: [10 Marks]

(a) lfa < Minimax(x) < 3, then ALPHABETA(, a, B) = Minimax(z)
(b) IF Minimax(x) <a, then ALPHABETA(z, a, B) < a
(c) If Minimax (x) B, then ALPHABETA(z, a, B)> B
2. Consider a d-ply fiull game tree with each node (except leaves) having a branching factor of h.
Prove or disprove the following:
Atleast bl +61LEAF nodes must be eualuated to find a finite Lower bound d finite upper
baund of the minimat value of the rvot node. O Marks]
3. Solve N-queens problenm using hill climbing. Define the statespace, neighbourhood function, and
-the evaluation funetion. Show the evolution of the algorithm for N'= 4. (10 Marks]
4. Using Particle Swarm Optimisation (PS0) minimize the function
min, f(z) = +
,yER

The parameters are swarm-size num-iterations 3,) inertia-weight (w] =0.7.

learning-coefficients løp, ol A8, upper-bound|bu]75. lower-bound|bp] =-5. For
randomness please use any value from (0, 1). For the initial positions of the suara elements and
their velocitics refer tothe table below: [10 Marks)
Particle Position(z, y) Velocity (va, Vy)
(2, 1) (0.0)
2 (-1,3) (0. 0)
3 (0, -2) (0.0)

6
 Aihcal ntelhgene- Endsen

\Mar 2. 2024

Total Marks: 50 Answer all the quest ions

(a) Dtinc consistent and admissible heuristics
b) Consder EIGHT-PUZZLE whh is defned as follows: In this gae. there is a square block
ontaining &cels with one emptv cell. The ain is to rearrange the cells into a specifc goal
mnhguration by slhding them intoadjarent empty spaces. Theobjective is to achiceve this
goal configuration from a given initial st ate using as few moves as possible. For this purpose
consider the heur ist ic function given brthe following
Count the nunber of tles that are not at their respective goal location.
Is this heuristic function consistent and admissible? Justifv.
5+2 Marks)

12 3 1 2 3 1 2 3

5 4 | 4 4

6 7 a
start state goal state

2. Bescibe vour solution to one of the assigment questions you provided. 5 Marks]
3. =sing resolution and rules of inference. show that the formula -A ’ ((A ’ B) A(4 ’ -B) is
a tautolog. (10 Marks]
4 What is the objective that SSS* seeks to achieve ? What is the nature of the priority queue
uSed in SSS ? Does ninimax and SSS" provide the same solution? Explain. How do you
justify that the operation performed in aSOLVEDmin node is valid? (2+1+2+5 Marks)
5. Consider the sliding-tile puzzle FROGS & TOADS. A tile can either slide or jump if the target
location is ernpty. The goal is to exchange the sets of black and white tles in the minimum
number of moves. Gie a state space characterization of the puzzle and propose an approach to
solve the problem. [8 Marks]

O00
O0C

6/ How can one develop strategies for the MAX player corresponding to the minimax. -3pruning
and SSS* algorithms. (1+2+2 Marks)
7. Consider the garne DOTS & LINES which is defined as follows: You are given aspeciñc lavout
of dots on a2D plane as shown in figure. Players take tuns connecting adjacent dots, The îst
player to omplete a triangle wins. Now, create he minimax tree for this game. ||5 Marks]
 S. For positive weiglht setting, DFS is not optimal. However iterative deepening DFS is optimal.
Agree or disagree? Argue. What is the advantage of iterative deepening A* over A*. (2 Marks]
9. Let the goal be located at level k of the search tree. Also let n * and nËDA+ be the number of
expansions performed by A* and IDA* respectively. Then show that (5 Marks)
26
nIDA* 262
b-1 b-1' where b is the branching factor

Table 1.5 Basic rules for derivations

Premise Conclusion Name

Gis in F FtG Assumption

FEG and Fc F F'tG Monotonicity
FtG Double ncgation
FEF,FFG FE (FAG) A-lutroduction
FE (FA C) FtF A-Elimination
Ft (FA G) FE (GA F) -Symmctry
FF FE (FV G) v-Introduction
FE (FvG).
FU{P}t H, FU{G) H F H v-Elimination
FE (FvG) FE (Gv F) v-Symmetry
FUF}FG FE(F-G) -Introduction
FE(F - G), FE F FEG --Elimination
FEF FE (F) ()-Introduction
FF(F) FEF (.)-Elimination
FE ((FG)AH) FE (FAGAH) A-Parenthescs rule
FE((FvG) vH) FE (FvGvH) v-Parent heses rule

Table 1.6 Morc rules for derivations

Rules
Name

F (Ev C)if and only if Ft (¬FA ¬G) v-Definit ion

FE(F- G) if and only if FEGEvG) -Definit ion
FE (F G) if and only if both F E(F- G) and F E(G-F)
--Definition
Predure ID\Driver
Input: Implicit problem graph with start node s, weight function w, heuristic h,
successor generation function Expand, and goal predicate Goal
Output: Path froms to teT, or 9 if no such path exists
; Initialize global threshold
bestPath - : Initialize solution path
while(bestPath = #and U' oo) : Goal not found, unexplored nodes left
:: Reset global threshold
: Initialize new global threshold
bestPath +IDA* (s.0.U) ; Invoke Alg. 5.8 at s
return bestPath : Terminate with solution path

Algorithm 5.7
Driver loop for IDA'.

Procedure IDA*
Input: Node u, path length g, upper bound U
Output: Shortest path to a goal nodeteT, or 9if no such path exists
Side effects: Update of threshold U'
: Terminate search
if (Goal(u) return Pathu)
Succ(u) + Expandu) : Generate successor set
for each vin Succ(u) : For all successors
if (g+w(u. v) +h(v) > U) ;: Cost exceeds old bound
if (g + wu, v) + h(v) < U) ; Cost smaller than new bound
U gw(u, v)+ h(v) ; Update new bound
else
:f-value below current threshold
p +IDA"(v.g + w(u, v), U) : Recursive call
if (p #Ø) return (u.p) ; Solution found
return
:: No solution exists

Algorithm 5.8
The IDA algorithm

min

mol
 8
h
89

() Ay) A (a