Hierarchiczne deglomeracyjne sieci SOM w analizie skupień
The hierarchical divisive SOM in the cluster analysis

Summary/Abstract: In the empirical studies hierarchical structure of units and clusters can be a problem. One solution to this problem is to build a hierarchical divisive SOM network. Two approaches can be distinguished: static and dynamic. The construction of such a network is hierarchical, as parts of the network layer of one network are the nucleus in the next layer. As a result, the net-work can learn much faster, the number of dead neurons will be reduced and much more detailed identification of the group structure will be possible. The aim of this paper is to analyse the prop-erties of divisive HdSOM network in the cluster analysis.

• Details
• Contents