Paper Title
Improvement Of Self-Organizing Maps Algorithm With Weighting Optimization
Abstract
The Self-Organizing Map (SOM) is a neural network algorithm based on unsupervised learning. It used for high
dimensional data visualization. In other words, this algorithm maps high dimensional data to low dimensions space. Weight
initializing is one of the main steps in SOM algorithm, because the proper initializing the weights has great influence on final
convergence of network and it guides convergence toward local or global minimum. In order to reduce the iteration number
and increase the rate of algorithm, we have decided to improve the initializing step of weights. For this purpose, weight
initializing phase spilt to two steps and as observed from results, iteration no. decreased significantly.
Keywords- Clustering, Self-Organizing Maps, Weighting Optimization.