How do I interpret the output of a trained SOM for classification.

I have a SOM network that I have trained on the histogram of grey scale images resized to 50x50. However, I am not sure how to use that network to classify other pictures that I expose to the network. How do i know which picture activated a particular neuron?

Answer 1

[Greg Heath]

Kohonen's advice: Convert the SOM to a LVQ neural network.

Hope this helps.

Thank you for officially accepting my answer.

Greg

[Romaine Carter]

Thank you, I came to a similar conclusion after a few minutes of reading, it seems SOM's are useful only for visualizing the classification, while LVQ's are what would do the actual classification.

Answer 2

[aarti birla]

Can you please share code for classification using SOM and LVQ?