Understanding high-dimensional real world data usually requires learning the structure of the data space. The structure may contain high-dimensional clusters that are related in complex ways. Methods such as merge clustering and self-organizing maps are designed to aid the visualization and interpretation of such data. However, these methods often fail to capture critical structural properties of the input. Although self-organizing maps capture high-dimensional topology, they do not represent cluster boundaries or discontinuities. Merge clustering extracts clusters, but it does not capture local or global topology. This paper proposes an algorithm that combines the topology-preserving characteristics of self-organizing maps with a flexible, adaptive structure that learns the cluster boundaries in the data.

PDF, PS

In Machine Learning: Proceedings of the 12th Annual Conference, Armand Prieditis and Stuart Russell (Eds.), pp. 55-63, Austin, TX 1995. San Francisco, CA: Morgan Kaufmann. 55-63. Technical Report AI95-238.

@MastersThesis{blackmore:ms95, title={Visualizing High-Dimensional Structure With The Incremental Grid Growing Neural Network}, author={Justine Blackmore and Risto Miikkulainen}, booktitle={Machine Learning: Proceedings of the 12th Annual Conference}, editor={Armand Prieditis and Stuart Russell}, school={Department of Computer Sciences, The University of Texas at Austin}, address={Austin, TX}, publisher={San Francisco, CA: Morgan Kaufmann}, pages={55-63}, note={Technical Report AI95-238}, url="http://www.cs.utexas.edu/users/ai-lab?blackmore:icml95", year={1995} }