The Growing Hierarchical Self-Organizing Map: Exploratory Data Analysis of High-Dimensional Data

Authors: 
Andreas Rauber
Wolfdieter Merkl
Michael Dittenbach
Type: 
Journal article
Proceedings: 
Publisher: 
IEEE Transactions on Neural Networks, 13-6
Pages: 
1331 - 1341
ISBN: 
Year: 
2002
Abstract: 
The self-organizing map is a very popular unsupervised neural network<br> model for the analysis of high-dimensional input data as in data mining<br> applications. However, at least two limitations have to be noted, which are<br> related, on the one hand, to the static architecture of this model, as well<br> as, on the other hand, to the limited capabilities for the representation<br> of hierarchical relations of the data.<br> <br> With our novel growing hierarchical self-organizing map presented in this <br> paper we address both limitations.<br> The growing hierarchical som is an artificial neural network model with <br> hierarchical architecture composed of independent growing self-organizing <br> maps.<br> The motivation was to provide a model that adapts its architecture during <br> its unsupervised training process according to the particular requirements <br> of the input data.<br> Furthermore, by providing a global orientation of the independently growing <br> maps in the individual layers of the hierarchy, navigation across branches <br> is facilitated.<br> The benefits of this novel neural network are first, a problem-dependent <br> architecture, and second, the intuitive representation of hierarchical <br> relations in the data. This is especially appealing in explorative data <br> mining applications, allowing the inherent structure of the data to unfold <br> in a highly intuitive fashion.<br> <br> Keywords: Self-Organizing Map (SOM), Data Mining, Hierarchical <br> Clustering, Exploratory Data Analysis, Pattern Recognition.<br> <br>
TU Focus: 
Information and Communication Technology
Reference: 

A. Rauber, W. Merkl, M. Dittenbach:
"The Growing Hierarchical Self-Organizing Map: Exploratory Data Analysis of High-Dimensional Data";
IEEE Transactions on Neural Networks, 13 (2002), 6; S. 1331 - 1341.

Zusätzliche Informationen

Last changed: 
22.01.2003 13:59:06
TU Id: 
136954
Accepted: 
Accepted
Invited: 
Department Focus: 
Business Informatics
Abstract German: 
Author List: 
A. Rauber, W. Merkl, M. Dittenbach