This is an example of a Kohonen's SOM that I wrote to study the actual working of it programmatically.
These instructions will allow you to fully understand the program.
What things you need to run this program
python2.7
numpy
The program is written for a predefined input set.
The training data used is also hardcoded into the program as this is just to understand the working of the SOM.
The number of output clusters is set to 2 in this particular program.
Even though the tag no_of_iterations is given as 100, this program currently runs only for 5 iterations/epochs.
I can update and explain the readme more as per questions asked. Feel free to get in touch at [email protected]