Data Clustering using Artificial Neural Network, Rough set Theory and Optimization Techniques

Artificial Neural Network in particular Self Organizing Map (SOM) has been widely used in clustering analysis operations. SOM maps high dimensional data space into two dimensions colored grid. The crisp clustering which employs one threshold to determine cluster boundaries has poor performance in many complex and high dimension data sets. In this paper to improve the performance of clustering analysis, SOM is used to reduce the data dimensionality and to have optimized cluster boundaries Rough set theory and Optimization techniques has been employed in the second stage. In this proposed approach, first using SOM to cluster and reduce the dimensionality of the data set and using Rough set theory and optimization problems to reduce uncertainty that involved in cluster analysis, in the second stage, seems that has much better performance in compare with crisp clustering methods.