Hierarchical Growing Neural Gas
This paper describes TreeGNG, a top-down unsupervised learning method that produces hierarchical classification schemes. TreeGNG is an extension to the Growing Neural Gas algorithm that maintains a time history of the learned topological mapping. TreeGNG is able to correct poor decisions made during the early phases of the construction of the tree, and provides the novel ability to influence the general shape and form of the learned hierarchy.
KeywordsTree Generation Delaunay Triangulation Graph Splitting Poor Decision Fuzzy ARTMAP
Unable to display preview. Download preview PDF.
- Everitt, B. (1993), Cluster Analysis, Edward Arnold, London, 3rd editionGoogle Scholar
- Hertz, J., Krogh, A., Palmer, R. (1991), Introduction to the Theory of Neural Computation, Addison-Wesley, Redwood City, CAGoogle Scholar
- Fritzke, B. (1996), Growing Self-Organizing Networks-Why?, In: Proc. European Symposium on Artificial Neural Networks, pp. 61–72Google Scholar
- Fritzke, B. (1995), A Growing Neural Gas Network Learns Topologies, Advances in Neural Information Processing Systems, pp. 625–632Google Scholar
- Köhle, M., Merkl, D. (1996), Visualising Similarities in High Dimensional Input Spaces with a Growing and Splitting Neural Network, Lecture Notes in Computer Science, 1112:581–586Google Scholar
- Burzevski, V., Mohan, C. K. (1996), Hierarchical Growing Cell Structures, In: Proc. IEEE Int. Conf. Neural Networks, Washington D.C., volume 3, pp. 1658–1663Google Scholar
- Cao, X., Suganthan, P. (2002), Hierarchical Overlapped Growing Neural Gas Networks with Applications to Video Shot Detection and Motion Characteristics, In: Proc. Int. Joint Conf. Neural Networks, IEEE, Hawaii, USA, volume 2, pp. 1069–1074Google Scholar