First page Back Continue Last page Overview Graphics
SOM Training
Based on rewarding the winning node
This is a form of competitive learning
Winners weights are adjusted to be closer to the input vector
Why not equal?
- We want the output map to learn regions, not examples