Last year's final exam
has been posted as a study guide. Realize
there are some things we didn't cover that
are on this final, and some things we did
cover that aren't on this final. Also,
a study guide
for this year's exam is now posted.
Homework 5 04/07/2003
Homework 5 has been posted and is
due on Monday, April 14, 2003.
Resources 03/24/2003
Added two pages containing Java and Prolog
resources.
Homework 3 02/28/2003
Homework 3 has been graded. I'm considering
a curve on this one, which I'll announce
in class on Monday.
Homework 2 02/27/2003
Homework 2 has been graded and is available
on my office door. Homework 3 will be available
tomorrow.
My Solution 02/26/2003
Here's my solution
to the URL search problem. You can try
whatever search techniques you want
and see how things work out. To prevent
hanging on slow connections, use the
option -Dsun.net.client.defaultConnectTimeout=[ms] with the Java interpreter, replacing [ms] with the number of milliseconds you're willing to wait for a connection.
Previous Midterm 02/26/2003
The midterm
from last semester has been posted.
A few notes, last year we didn't cover
adversarial search, so we got further
in the material, by getting to first-order
logic. We haven't covered that yet this semester.
Also, there are Prolog questions which can
be ignored.
Finally, I'll be looking at other midterms
from other instructors to get new questions
this year. Look at the AIMA web site
for links to those exams.
Homework 4 02/24/2003
Homework 4 has been posted.
It's due Monday, March 24, 2003.
Problem 4.8 02/18/2003 Final corrected interpretation of problem 4.8!The problem description is very vague if you didn't notice. I apologize for the confusion on this one. Here's the final correction to how the problem should be interpreted. Assume, that you have a partial tour T. To estimate the cost of completing the tour, you calculate s(n) the shortest distance from the start of T to one of the remaining cities, e(n) the shortest distance from the end of T (the last city selected) to one of the remaining cities, and r(n) an estimate of how much it'll cost to travel through the remaining cities. So h(n) = s(n) + e(n) + r(n)
For the MST heuristic, the estimate of how
much it'll cost for the remaining cities
is the MST cost of the remaining in cities.
That is, r(n)= MST cost of remaining cities
For the SLD heuristic, the estimate of
how much it'll cost to tour the
remaining k cities is the total of the
shortest k-1 distances between the
remaining cities. That is, r(n)=sum of
smallest k-1 distances between remaining
cities.
Problem 4.8 02/18/2003 Corrected interpretation of problem 4.8!
For the SLD heuristic, total the SLDs from
the "last" city in the partial tour to all
remaining cities.
For the MST heuristic, calculate the MST
cost for the cities not contained
in the partial tour.
Search Code 02/17/2003
The search code has been posted. Get it
from the homework page.
Search Code 02/17/2003
I'll be posting the first version of the
search code today with an example program.
Due to my delay in getting the search code
posted, I'm extending the due date
of the final problem on homework 3
to Monday, Feb. 24. This is the for the
final problem only.
Homework 02/10/2003
Homework 3 has been posted. It's due
Wednesday, Feb. 19.
Agent Simulator 02/03/2003
I updated the the agent simulator for better random number handling. Subclasses can override the random instance variable for random number generation. I believe this might still not handle the random numbers to generate the initial width and height of the environment. I'll look into things a little later. In the mean time, you can copy the code from the GridEnvironment no parameter constructor, setting random to your own Random object before executing that code.
Agent Simulator 02/03/2003
The next release of the agent simulator is now available. This simulator implements a performance measuring environment for the monkey and banana. The simulator shell displays scores for agents in the environment. The monkey now eats the banana, and you can get the contents of a location if the environment allows it.
I've changed the simulator shell a
little bit. Now, you should create a
subclass of
cosc159.simulator.Simulator and
create a main method that
instantiates the simulator and the
environment used for simulation. See
cosc159/simulator/MBSimulator.java
for details. As a result of this change,
to run the monkey and banana simulation,
use make mbrun or java -cp
classpath
cosc159.simulator.MBSimulator
Compilation 01/27/2003 A
few things about the simulator. First, if
you install cygwin be sure to install the
make package which isn't installed by
default. I typically install everything,
which requires you to click the two
circling arrows in the cygwin setup box
(it takes a while to update after you
click them, so be patient).
If you don't want to use cygwin under
Windows, you can set your CLASSPATH
variable to contain the DOS directory
containing the simulator, the
jlfgr-1_0.jar file, and your current
directory. For example,
set CLASSPATH=C:\AgentSimulator;C:\AgentSimulator\jlfgr-1_0.jar;.
COSC 159: Fundamentals of Artificial Intelligence - Call# 65468 - Spring 2003
Last year's final exam
has been posted as a study guide. Realize
there are some things we didn't cover that
are on this final, and some things we did
cover that aren't on this final.
Homework 5 04/07/2003
Homework 5 has been posted and is
due on Monday, April 14, 2003.
Resources 03/24/2003
Added two pages containing Java and Prolog
resources.
Homework 3 02/28/2003
Homework 3 has been graded. I'm considering
a curve on this one, which I'll announce
in class on Monday.
Homework 2 02/27/2003
Homework 2 has been graded and is available
on my office door. Homework 3 will be available
tomorrow.
My Solution 02/26/2003
Here's my solution
to the URL search problem. You can try
whatever search techniques you want
and see how things work out. To prevent
hanging on slow connections, use the
option -Dsun.net.client.defaultConnectTimeout=[ms] with the Java interpreter, replacing [ms] with the number of milliseconds you're willing to wait for a connection.
Previous Midterm 02/26/2003
The midterm
from last semester has been posted.
A few notes, last year we didn't cover
adversarial search, so we got further
in the material, by getting to first-order
logic. We haven't covered that yet this semester.
Also, there are Prolog questions which can
be ignored.
Finally, I'll be looking at other midterms
from other instructors to get new questions
this year. Look at the AIMA web site
for links to those exams.
Homework 4 02/24/2003
Homework 4 has been posted.
It's due Monday, March 24, 2003.
Problem 4.8 02/18/2003 Final corrected interpretation of problem 4.8!The problem description is very vague if you didn't notice. I apologize for the confusion on this one. Here's the final correction to how the problem should be interpreted. Assume, that you have a partial tour T. To estimate the cost of completing the tour, you calculate s(n) the shortest distance from the start of T to one of the remaining cities, e(n) the shortest distance from the end of T (the last city selected) to one of the remaining cities, and r(n) an estimate of how much it'll cost to travel through the remaining cities. So h(n) = s(n) + e(n) + r(n)
For the MST heuristic, the estimate of how
much it'll cost for the remaining cities
is the MST cost of the remaining in cities.
That is, r(n)= MST cost of remaining cities
For the SLD heuristic, the estimate of
how much it'll cost to tour the
remaining k cities is the total of the
shortest k-1 distances between the
remaining cities. That is, r(n)=sum of
smallest k-1 distances between remaining
cities.
Problem 4.8 02/18/2003 Corrected interpretation of problem 4.8!
For the SLD heuristic, total the SLDs from
the "last" city in the partial tour to all
remaining cities.
For the MST heuristic, calculate the MST
cost for the cities not contained
in the partial tour.
Search Code 02/17/2003
The search code has been posted. Get it
from the homework page.
Search Code 02/17/2003
I'll be posting the first version of the
search code today with an example program.
Due to my delay in getting the search code
posted, I'm extending the due date
of the final problem on homework 3
to Monday, Feb. 24. This is the for the
final problem only.
Homework 02/10/2003
Homework 3 has been posted. It's due
Wednesday, Feb. 19.
Agent Simulator 02/03/2003
I updated the the agent simulator for better random number handling. Subclasses can override the random instance variable for random number generation. I believe this might still not handle the random numbers to generate the initial width and height of the environment. I'll look into things a little later. In the mean time, you can copy the code from the GridEnvironment no parameter constructor, setting random to your own Random object before executing that code.
Agent Simulator 02/03/2003
The next release of the agent simulator is now available. This simulator implements a performance measuring environment for the monkey and banana. The simulator shell displays scores for agents in the environment. The monkey now eats the banana, and you can get the contents of a location if the environment allows it.
I've changed the simulator shell a
little bit. Now, you should create a
subclass of
cosc159.simulator.Simulator and
create a main method that
instantiates the simulator and the
environment used for simulation. See
cosc159/simulator/MBSimulator.java
for details. As a result of this change,
to run the monkey and banana simulation,
use make mbrun or java -cp
classpath
cosc159.simulator.MBSimulator
Compilation 01/27/2003 A
few things about the simulator. First, if
you install cygwin be sure to install the
make package which isn't installed by
default. I typically install everything,
which requires you to click the two
circling arrows in the cygwin setup box
(it takes a while to update after you
click them, so be patient).
If you don't want to use cygwin under
Windows, you can set your CLASSPATH
variable to contain the DOS directory
containing the simulator, the
jlfgr-1_0.jar file, and your current
directory. For example,
set CLASSPATH=C:\AgentSimulator;C:\AgentSimulator\jlfgr-1_0.jar;.
After that, change into each subdirectory
of cosc159 and execute
javac *.java
Then you can execute the simulator by
changing back to the AgentSimulator directory
and executing
java cosc159.simulator.Simulator
If someone wants to write a DOS batch file
for me, I'll include it in the next
update of the simulator.
Simulator
Code 01/24/2003 A
prerelease
version of the agent simulator code is
now available. It's mostly working, but
has a few outstanding issues. To build
and run, you'll need to use studsys, some
other Unix based machine (with GNU make
installed) or install cygwin
on a Windows machine. Edit the file
Rules.mk and modify the
PROJDIR environment variable to
the location of the top level directory of
the simulator. Type the command
make to build everything, and the
command make run to run a test
simulator. Single step, pause, and run
all work. The other buttons do not.
You should spend some time looking
through the code to see what I've done.
In particular, pay attention to the
monkey and banana related code, because
that represents pretty much what you'll
need to implement for your own simulation.
Bring questions to class on Monday.
News
Article 01/15/2003 I
found this news
article, which discusses the use of AI
to produce games with an infinite number
of story lines.
Homework 1 01/15/2003
Homework 1 has been posted. It's due
Wednesday, January 22.
Welcome 01/13/2003
Welcome to the home page for Dr. Struble's
Spring 2003 COSC 159: Fundamentals of
Artificial Intelligence at Marquette
University.