This can accomplished quantizing the weight vector into four bins

C++ Neural Networks and Fuzzy Logic:Application to Pattern Recognition

	C++ Neural Networks and Fuzzy Logic by Valluru B. Rao M&T Books, IDG Books Worldwide, Inc.

In this chapter, you will use the Kohonen program developed in Chapter 11 to recognize patterns. You will modify the Kohonen program for the display of patterns.

An Example Problem: Character Recognition

The letter A is represented by the values:

0 0 1 0 0
0 1 0 1 0
1 0 0 0 1

For the characters A and X you would end up with the following entries in the input file, input.dat:

0 0 1 0 0 0 1 0 1 0 1 0 0 0 1 1 0 0 0 1 1 1 1 1 1 1 0 0 0 1 1 0 0 0 1

We will present the Kohonen map with many such characters and find the response in output. You will be able to watch the Kohonen map as it goes through its cycles and learns the input patterns. At the same time, you should be able to watch the weight vectors for the winner neurons to see the pattern that is developing in the weights. Remember that for a Kohonen map the weight vectors tend to become aligned with the input vectors. So after a while, you will notice that the weight vector for the input will resemble the input pattern that you are categorizing.

Representing the Weight Vector

0 < weight <= 0.25 Light-dotted rectangle

0.25 < weight <= 0.50 Medium-dotted rectangle

Table 12.2ASCII Values for Rectangle Graphic Characters