Download Cognitive Modeling - Artificial Intelligence - Exam and more Exams Artificial Intelligence in PDF only on Docsity!
C S C 4 8 0 A R T I F I C I A L
I N T E L L I G E N C E
M I D T E R M E X A M
SECTION 1
PRO F. FRA N Z J. KU RF ES S
CAL PO L Y, COM P UTE R SC IEN CE DE PA RT MEN T
This is the midterm exam for the CSC 480 Artificial Intelligence class. You may use textbooks, course notes, or other material, but you must formulate the text for your answers yourself. The use of calculators or computers is allowed for numerical calculations, but not for the execution of algorithms or programs to compute solutions for exam questions. The exam time is 80 minutes.
Student Name: Student ID:
Signature: Date:
PA RT 1: MU LTIP LE CH OI CE QUE S TIO NS
Mark the answer you think is correct. Unless otherwise noted, there is only one correct answer. Each question is worth 2 points, for a total of 20 in Part 1. a) Which statement is the best characterization of cognitive modeling?
❑ The formulation of algorithmic descriptions of building blocks for intelligent
systems.
❑ The formal specification of abstract reasoning mechanisms for systems that
represent and manipulate knowledge
❑ The construction of systems that exhibit behaviors necessary for solving tasks
requiring intelligence.
❑ An attempt to describe the way the human mind functions.
b) Which event is considered the "birth" of the field of Artificial Intelligence?
❑ The formulation of the Turing test by Alan Turing in 1950.
❑ A workshop in the summer of 1956 at Dartmouth.
❑ The development of the Lisp programming language in 1958.
❑ The victory of the Deep Blue computer system over the chess world champion,
Gary Kasparov, in 1998. c) In which of the following areas is human intelligence still significantly superior to artificial intelligence?
❑ Exact retrieval of patterns like strings from large sets of examples (e.g. in the index
of a search engine).
❑ Planning and scheduling of actions for intricate tasks (e.g. the assembly of
automobiles).
❑ Identifying solutions to a problem from a collection of previous cases that describe
solutions to similar problems (e.g. in a help desk application).
❑ Translation from one natural language into another.
d) What is an ideal rational agent?
❑ Conceptually, and all-knowing agent that could predict the outcome of any action.
❑ An agent that maximizes its performance metric for a given task, percept sequence,
background knowledge, and configuration of the environment.
❑ An agent that can explain its choice of an action.
❑ An agent that is capable of predicting the actual outcome of an action.
e) In general, which of the following environments is the least challenging for agents?
❑ fully accessible, discrete, episodic, static
❑ static, non-episodic, continuous, partially accessible
❑ inaccessible, episodic, dynamic, continuous
❑ deterministic, non-episodic, dynamic, continuous
PA RT 2: SHO RT QUE S TIO N S
In this part of the exam, you must answer the questions in one or two paragraphs. Each question is worth 5 or 10 points, for a total of 30 in Part 2.
- Explain the difference between an ideal rational agent and an omniscient agent. [10 points]
- Give a short PAGE description for an agent residing in a personal computer system, with the task of identifying documents with similar contents. [10 points] Aspect Description P A G E
PA RT 3: MA ZE SEA RCH
In this scenario, an agent is trying to traverse a maze from the starting point S to the goal point G. At each step, the agent can move in one of the four compass directions; each moves, independent of the direction, costs the agent one cost unit. The agent always considers alternative moves in the following order:
- Move North
- Move East
- Move South
- Move West In the following parts, you need to apply different search algorithms to solve this navigation problem. Number the squares in the order the agents visits the squares, starting with 0 at the starting point. You do not need to re-expand nodes already visited; this means that you can “jump” from the current node to the next node in the queue. In those algorithms that use it, calculate the path cost on the basis of one cost unit per move. The heuristics to use in the respective algorithms is the difference between the horizontal position of the current node and the goal node, plus the difference in the vertical position of the current node [xn,yn] and the goal node [xg,yg], adjusted by a small value to break symmetries: h([x,y]) = (|xg - xn|) + 0.99(|yg - yn|) So, for the starting point, node [1,3], the heuristics with respect to the goal, node [6,5], is (6-1) + 0.99*(5-3) = 5 + 1.98 = 6.98. For the following algorithms, you need to do the following tasks
- Mark the sequence in which the nodes are visited in the maze. You can do this directly in the copy of the maze for each algorithm.
- Draw the corresponding search tree. It might be advisable to draw the complete search tree first on a separate sheet of paper, and then draw the relevant parts next to / below the maze.
- Fill in the table with the information about the search trace. The size of the table does not necessarily relate to the number of steps in the algorithm. Extend the table if necessary.
G
S
0 1 2 3 4 5 6 7 8 9 0 9 1 2 3 4 5 6 7 8 a) Traverse the maze from the starting point S to the goal G according to the Depth-First Search method. [15 points] Step Current Node Path Cost Heuris- tic F-Cost Queue 0 S 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G
S
0 9 1 2 3 4 5 6 7 8 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 c) Traverse the maze from the starting point S to the goal G according to the A* Search method. Use the simple addition f(n) = g(n) + h(n) to calculate the f-cost for the A* algorithm, with path cost and heuristic as defined earlier. [17 points]
Step Current Node Path Cost Heuris- tic F-Cost Queue 0 S 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
