2

I roughly know SOM and it is mapping its network to different clusters of trained data. How to implement to count the number of clusters using SOM? I use KNNL library here to implement SOM. In its demo, it is shown how to train and test only. How can I implement it for counting number of clusters? I know I can also use DBSCAN for cluster counting as well. But first, I like to implement SOM for cluster counting. My input data is 2D data representing points in 2D space like 

(132.181,0.683431),
(136.886,0.988517),
(137.316,0.504297),
(133.653,0.602269),
Brian Tompsett - 汤莱恩
  • 5,753
  • 72
  • 57
  • 129
batuman
  • 7,066
  • 26
  • 107
  • 229
  • You might find this answer usefull, http://stackoverflow.com/questions/13074930/identify-clusters-in-som-self-organizing-map/13089569#13089569 – pater Feb 18 '14 at 09:37

0 Answers0