Basic function of the training algorithm of the traditional SOM neural network is to cluster the Weight Vector of neurons of the SOM output layers into separate clusters. The training algorithm of the traditional SOM neural network is as follows:
Step 1: Initializing randomly weight of neurons of the SOM output layer and assign value to NC(t) (the radius of the neighbor area). Assign value 1 to cycle t.
Step 2: Presenting an input vector (t) to SOM input layer and normalize v(t) then calculate the Euclidean distance from input vector v (t) to all the vectors of all neurons of the SOM output layer and choose neuron with the minimal Euclidean distance from input vector v (t) in to the corresponding neuron.
[…]