Kohonen update rule
From Piki
The Kohonen update rule is a competitive learning update rule used to train Self-Organizing Maps (SOMs). It is named after Teuvo Kohonen, the inventor of the SOM.
| Kohonen | |
| Family | Competitive |
| Direction | Forward |
| Deployable | Yes |
| Supervised | No |
Contents |
Usage
Main article: Self-organizing map
The Kohonen update rule is closely related to other competitive rules, such as the Neural gas update rule. The difference between a neural gas and a Self-Organizing Map is that the map has elastic links between the neurons while the gas neurons are free to move unrestricted.
Algorithm
Weight update:
where 
are the weights, 
the input vector, 
the learning rate and 
the neighbourhood function.
Settings
The settings can be modified using the settings browser.
| Kohonen update rule settings |
|---|
|
|
See also
- Update rule - General article on update rules.
- List of update rules - List of all update rules.
- Competitive learning - General article on competitive learning.
- Self-organizing map - General article on self-organizing maps.
- SoMap layer block - Synapse Block that uses this rule.
- SOM View visualizer - Synapse Visualizer that uses this rule.