Growing Multi-Dimensional Self-Organizing Maps for Motion Detection
Udo Seiffert · Studies in fuzziness and soft computing · 2002
The standard Self-Organizing Map consists of a two-dimensional rectangular grid of neurons. For many applications this represents a very good target to reduce the dimensionality of the input data. However, occasionally a multi-dimensional layer, keeping more than two dimensions of the input data, might be more advantageous. This sometimes also called hypercube topology can be considered as the universal case of the standard topology. This chapter gives an introduction and demonstrates basic properties by means of applications from motion picture analysis. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.