Self-Organizing Map
Feature vectors are input to SOM
Value (position) of each SOM node randomly initialized
For each input vector, find closest node (Euclidean distance)
Adjust weight of winning node so will be even closer to vector (more likely to be picked next time)