Neural Networks


Class Syllabus

April 11 2005

Please go to the following web site:

Interesting ANN site

Click on the "Associative Memories"

Learn first to run the program, then look at the code. Remember, it is Javascript not
Java, so there will be some learning curve. You can use the three preset patterns as well
create your own memories.

Carefully examine the Javascript implementation of this BAM (Bidirectional Associative Memory)
so that you can understand the processing involved. You will be writing a similar BAM program 
in Java. BTW looks like we do not need to use the bipolar value when we apply the inputs 
to the the weights. Recall that was the question in class (4/13/2005). We do need the bipolar 
input values when we set the weights.

April 18 2005

You are to ivestigate a variation on ANNs called the Boltzmann Machine and a technique called 
"simulated annealing". Write a one-page tutorial on the topics, and hand it in with your exam
on Wednesday. You should include a couple examples of applications. You can use the web to 
research, but be sure the resultng work is your own words throughout.

April 20 2005

In class Exam


Programming Assignments

Program 1 - GA 3-Color Graph problem

Program 2 - Perceptron

Program 3 - 2-Level Backpropagation LCD Display

Program 4 - Hopfield TSP Recurrent Network

Program 5 - Bidirectional Associative Memory (BAM) ANN


Class Handouts

Neuron Demo1 - Source Code

Neuron Demo2 - Source Code (use WinZip)

TSP Demo - Source Code (use WinZip*)

Graph Draw Demo - Source Code

* save the jar file, then use WinZip
page footer