Title :
Topology preservation in self-organizing maps
Author :
Kiviluoto, Kimmo
Author_Institution :
Neural Networks Res. Centre, Helsinki Univ. of Technol., Espoo, Finland
Abstract :
This paper concentrates on the following issues: (1) discussion on what kind of mapping is produced by the SOM algorithm; (2) introduction of a quantitative measure of continuity for the mapping produced by SOM; (3) introduction of a variant of SOM, called the AdSOM, with locally adapting neighborhood radii
Keywords :
self-organising feature maps; topology; AdSOM; locally adapting neighborhood radii; mapping; quantitative continuity measure; self-organizing maps; topology preservation; Biological neural networks; Information processing; Intelligent networks; Lattices; Network topology; Neural network hardware; Neural networks; Neurons; Probability distribution; Self organizing feature maps;
Conference_Titel :
Neural Networks, 1996., IEEE International Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-3210-5
DOI :
10.1109/ICNN.1996.548907
