Java程序辅导

Semester 1, 2003  Page 1 of 1 
B219 Intelligent Systems 
 
Laboratory 1 
 
Introduction 
 
 
Before you start your laboratory work: 
 
§ You will need your username and password to log into the system. If you do 
not have the username and password, please seek help from BITL Help Desk. 
 
§ All your own work should be kept on your disks (3.5inch, 1.44MB). However, 
on the machine, there is hard disk drive (c:). You can copy your files on to the 
hard disk and work off it rather than your own diskette for a faster response. 
But remember to save your work on to your own disk at the end of the session. 
Any files left on the hard disk may be deleted to create space if necessary. 
 
Software requirement for the Laboratory exercises: 
 
§ MS Windows Operating System 
§ Web Browser Software 
§ JAVA SDK (http://java.sun.com/) 
§ JavaNNS (http://www-ra.informatik.uni-tuebingen.de/downloads/JavaNNS/) 
§ SNNS (http://www-ra.informatik.uni-tuebingen.de/SNNS/) 
§ Risk Analysis and Decision Making Software (Neuro3) 
(http://www.npto.doe.gov/Software/miscindx.html) 
§ ECC-Eliza V4.08 (http://ecceliza.cjb.net/) 
§ Amzi! Logic Explorer  
(http://www.amzi.com/download/amzi_logic_explorer.htm) 
 
 
In this lab, you will be introduced to Artificial Neural Network (ANN) by looking 
into the JavaNNS package and study the SNNS documentations. 
 
The Java Neural Network Simulator (JavaNNS) is a simulator used to understand the 
function of different artificial neural networks JavaNNS is actually a successor of 
Stuttgart Neural Network Simulator (SNNS) with Java GUI. It is based on SNNS 4.2 
Kernel with better graphical user interface. 
 
In order to run the JavaNNS, Java Runtime Environment (JDK) needs to be installed. 
JavaNNS has proved to work well with Java JDK 1.3. If you need to install JavaNNS 
in your home computer, do remember to download and installed JDK 1.3 available 
from http://java.sun.com/j2se/1.3/ before JavaNNS will work. 
 
Semester 1, 2003  Page 2 of 2 
To gain faster access to the SNNS user manual and JavaNNS user manual, you should 
access them from the local server. 
 
The URL for SNNS in the local server is  
http://www.it.murdoch.edu.au/~wong/b219/SNNS/ 
 
The URL for JavaNNS in the local server is 
http://www.it.murdoch.edu.au/~wong/b219/JavaNNS/ 
 
Please bring up the main page of the SNNS manual using any web browser, and click 
on the “The SNNS User Manual”. In this laboratory, look at the sections under 
“Neural Network Terminology” to gain an introductory understanding of the basic of 
ANN especially feedforward neural networks using backpropagation learning 
algorithm. 
 
Questions to be submitted at the end of the semester: 
(Please note that tutor will not give you the answers for the following questions) 
 
1. What are the three main units in ANN? Which are the units that are visible 
from outside world? 
2. What is generalisation of a neural network? 
 
Practical Exercises: 
 
1. Start JavaNNS program from the START menu while keeping your web 
browser running at the back. 
2. Open an example file using “FILE” then “OPEN” in JavaNNS. Click on the 
subdirectories “examples” and look for the file “letters_untrained.net”. 
3. Switch to your web browser, and click on “An Example of a simple Network”. 
4. Read through it and switch back to JavaNNS to see you can make sense out of 
it. 
5. You can also open the trained file with the file name “letters.net” to see the 
trained network. 
6. When JavaNNS asks “Load corresponding configuration file”, just click 
“Yes”. 
 
After all the above exercises, you should be able to gain an insight on how ANN is 
used to recognise letters.