Meaning of num_iteration inside train method about minisom

[JustGlowing]

hi Karim,

Please refer to the documentation: https://github.com/JustGlowing/minisom/blob/master/minisom.py#L395

To details about how num_iterations affects sigma and how the updates are performed, have a look at this method: https://github.com/JustGlowing/minisom/blob/master/minisom.py#L337

Regards

[drkj10]

Hi,

Thank you very much for your response.

I read the documentation but I need your confirmation please. It is not clear for me.

When use_epochs=False, we apply at each iteration before passing to the second row the following instruction. At the first iteration, we update the weights of all the points inside the circle whose radius is sigma. At the second iteration, the same instruction but using sigma(t) = sigma / (1 + t/T). At the end of num_iteration, we pass to the next row. In other words, each 'Best Matching Unit' is treated num_iteration times before passing to the next BMU.

Best regards.

Karim