Universität des Saarlandes
Fachrichtung 6.2 - Informatik
Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster
FR Informatik • Univ. d. Saarlandes • 66041 Saarbrücken
Final Exam
Introduction to the Methods of Artificial Intelligence Summer Term 2004
Family Name First Name Immatriculation No
Major Field of Study Minor Field of Study Diploma / Master
Date Signature
22.07.04
1. Terminological logics: Draw the following elements into the ontology that is depicted below:
a) subsumption relation(s), that can automatically be computed by the classifier.
b) partitions and coverings (if there are any).
Give reasons for your decisions in both cases (10 points).
2. Use the interpretation function to provide the modeltheoretic semantics for minimumnumberrestriction (atleast n R) (5 points)
3. Explain the difference between forward and backward chaining (Vorwärts und Rückwärtsverkettung) in production systems. (10 points)
4. RETEalgorithm: The following table represents a given memory status. The tokens are represented by their time stamps. Complete the depicted RETEnetwork by adding the contents of alpha and beta memory. (15 points)
time stamp class name color weight stands on
13 cuboid Q1 blue heavy bottom
36 cuboid Q2 red light bottom
56 pyramid P1 yellow light bottom
68 pyramid P2 white NIL Q1
129 cube W1 blue heavy bottom
223 cube W2 blue heavy NIL
5. Explain the syntax and the purpose of the OPS5 rule that is depicted below.
Assume that the element (FinalValue ^value 2004) is added to the database what does the output of the rule look like ? (10 points)
(P ruleBuilder
(FinalValue ^Value <x>) (BUILD \\ (GENATOM)
(Result ^Value \\ <x>)
--> (WRITE |final value reached|) (HALT) )
6. Bipolar Inheritance Networks: (10 points)
a) Provide the definition of the concatenation of paths (downward, coupled).
b) What are the constraints for a path being inheritable within a set of paths ?
7. Given a bipolar inheritance network like the one that is depicted below. Provide the credulous (leichtgläubige), the sceptical (skeptisch), and the ideally sceptical (ideal skeptische) extension(s).
(10 points)
C
A
B E D
F G
8. Frames (10 points):
Replace the place holder in the code below with the appropriate elements.
9. Propositional calculus: Given are the premisses i ... iv as well as the theorem v.
Prove or disprove that the theorem is implicated by the premisses.
Formal: AND(i,ii,iii,iv) > v. Use the resolution method. (15 points).
i) if it rains and the air is cold then the summer party is canceled.
ii) if the summer party is canceled then we are frustrated.
iii) it rains.
iv) the air is cold.
v) we are frustrated.
10. Given is the following problem: „A monkey is in a room that contains a chair and a bunch of bananas. The bananas hang from the ceiling so that the monkey cannot reach them. How can the monkey get the bananas ?“ (15 points)
The problem can be represented formally in the following way:
P(x,y,z,s) In state s the monkey is at location x. The bananas are at location y, and the chair is at location z.
R(s) The state in which the monkey reached the bananas.
P(A,B,C,S1) In the initial state S1, the monkey is at location A, the bananas are at location B, and the chair is at location C.
The following axioms are available:
the monkey can always go to the chair.
if the monkey is at the chair then it can carry the chair to the bananas if the monkey and the chair are at the place where the bananas are then the monkey can reach the bananas.
Show how this problem can be solved using Green's answer predicate. To do so, complete the following solution:
(C3) + (C5) heuristics: use the answer predicate as soon as possible
renaming of variables
11) DefaultLogic: Determine the extension(s) of the following default rule: (5 points)
W = {A}
D={A: B B , A:C
C ,B : D
D ,B :¬D∧¬C
¬D }
12) Define and explain the two performance measures (Leistungsmaße) for tree search techniques „penetrance“ (Penetranz) and „effective branching factor“
(Effektiver Verzweigungsfaktor) mentioned in the lecture. (5 points)
13) STRIPS: Show how the monkeybananaexample (see exercise 10) can be
represented so that STRIPS can generate a plan. Use the following actions: „go to chair“, „push chair under the bananas“, „climb on chair“, „grab the bananas“.
(10 points)
14) To process a sequence of robot actions, the following 'modules' are needed: a plan generator, a plan monitor, and a plan executor.
The execution of a plan can fail due to mechanical inaccuracies. We assume that no sensorical errors happen during the monitoring of the plan. (5 points)
When and how can triangle tables be applied ?
15) Given is an arbitrary triangle table. (5 points) a) What is represented by the first kernel ?
b) What happens if even the first kernel doesn't match when you seek the highest numbered matching kernel?
16) Modelbased diagnosis: See the following schematic diagram: (Addi are adders, Mi multipliers) (15 Points)
a) Define diagnosis.
b) Complete the table with the corresponding diagnoses after each measurement.
Measurements Diagnoses
a = 2, b = 1, c = 1 h = 9
i = 3 d = 3 f = 3 e = 1
17) ReasonMaintenanceSystems: (15 points)
a) Mark the following dependency network in a complete, consistent and well
founded manner.
b) Determine a supporting tree for node U1.
c) Determine the supporting sets for the nodes U2 and U13.
18) Image Understanding: Mark the junctions in the following geometric object and indicate the order in which you marked them. Line types are: <, >, +, (10 points)
19) Consider the following Bayesian network (15 points)
a) Compute the probability for event B, under the condition that the events C and E already occured but events A and D didn't. (conditional probability!)
b) Is the Dseperation criteria fulfilled, if evidence is given only for node C?
Explain.
20) Name the 4 kinds of inferences for Bayesian networks from the lecture and explain them briefly. (5 points)