A novel self-organizing map learning technique using community neuron on the map

Self-organizing map (SOM) is an artificial neural network tool that is based on unsupervised learning technique. This is used to produce a low dimensional representation of the input space, called a map. In conventional SOM, a winner is found. The weight vector of winner and its neighbors are updated. The learning of neighbor´s neurons is controlled by the distance from the winner on the map. The neurons which are closer to winner learns more. Due to this technique, few neurons become winner again and again. In this paper, we modify the learning technique using community neuron. The community neuron is found in 1- neighborhood of winner neuron. Several simulations are used to illustrate the effectiveness of the proposed algorithm. The learning capabilities are evaluated using three well known measurements, which are widely used to evaluate the performance of learning algorithms.