I've reviewed weeks worth of material and everyone stops at identifying the winning neuron and plotting the SOM.
However, how do I actually use this output?
Unsupervised SOMs are mainly used for clustering and visualization.
Clustering: Find datapoints which have their winning neuron close to each other. These datapoints share similarities and might belong to the same cluster or class.
Visualization: Plot the SOM grid and try to include the information about the winning neuron per datapoint into it. Examples are the World Poverty Map, RGB clustering, and this blobs example.