Animal Neuron Pattern Classification and Clustering using Self-Organizing Map (SOM) through Learning Rate and Minimal Cut

Authors have proposed a new neural classification model called Pattern Classification - self-organizing maps (PC-SOM), representing a best matching unit (BMU) for small SOM networks. Each vector of the input database is trained individually to provide best results by computing D(j) is minimum for one class only. Have considered one animal data set for classifying patterns as small animal or moderate animal or large animal by looking into the learning rate and minimal cut. This model is optimized in terms of number of epochs and number of iterations. The experimental results demonstrate that the proposed generalized algorithm PC-SOM outperforms for classifying the vectors in specific classes.