Practical Applications of Data Mining

6.9 learNiNg veCtor QuaNtizatioN (lvQ) model 253

weights as 0 or 1 between the hidden layer and the output layer. These val-

ues are called training pairs. In the learning process, if the training pair and the

hidden nodes are close enough, then set the weight between the inner node

and the input layer as 1, otherwise choose 0. In the hidden layer and the output

layer, the computed values are forced to be the same as the desired values by

changing the weights. The process described above is iterated as many times as

needed until the desired output is obtained.

Recall Process

Set the parameter.1.

Read in 2. W [ i][h] and W [ h][y].

Input the test data 3. X.

Calculate the output 4. Y.

Calculate the output 5. H in the hidden layer.

net[h]