Node Exchange for Improvement of SOM Learning

Abstract

Self Organizing Map (SOM) is a kind of neural networks, that learns the feature of input data thorough unsupervised and competitive neighborhood learning. In SOM learning algorithm, every connection weights in SOM feature map are initialized to random values to covers whole space of input data, however, this is also set nodes to random point of SOM feature map independently with data space. The move distance of output nodes increases and learning convergence becomes slow for this. To improve SOM learning speed, here I propose a new method, node exchange of initial SOM feature map, and a new measure of convergence, the average of the move distance of nodes. As a result of experiments, the average of the move distance of nodes comes to short that it becomes about 45%, and learning speed is improved that it becomes about 50% by this method.