Self-organizing maps (SOM) are useful for displaying and clustering high dimensional vectors on a two (or three) dimensional map. Below are a series of experiments I made using code originally taken and adapted form here.



map trained with a series of star shaped polygons, all nodes displaying current values



map clustering 1683 four letter words of the english language



map clustering a series of desktop icon, the dimensions are the colors of 16 (4×4) regions