Prototype Generation Using Self-Organizing Maps for Informativeness-Based Classifier

Algorithm 1

Prototype Generation based on SOM briefly described in a pseudo-code.

Input: The weight vectors of SOM Map trained (); training objects dataset (); and an unknown sample ()

Output: the label of unknown sample ()

 Compare the unknown sample with each weight vector of using Eq. (3)

 The units to be visited are defined by Eq. (4) and the input patterns objects are retrieved by Eq. (5), recovering a reduced dataset

 training .

 The reduced dataset and the unknown sample are used by a classifier (NN or NN) that return the object class.