Self-organizing map
Mapa autoorganizado
- Date
- 1982 · confidence high
- Place
- Espoo, Finland
- Credited to
- Teuvo Kohonen
- Field
- Knowledge & symbols
What it is
An unsupervised-learning neural network that reduces high-dimensional data to a low-dimensional, typically two-dimensional, map while preserving its topological structure. It trains via competitive learning rather than backpropagation, and is used to visualise and cluster complex data.
Introduced by Finnish professor Teuvo Kohonen in the 1980s; also known as a Kohonen map or network.
What it led to
- visualisation and clustering of high-dimensional data
- applications in data mining, pattern recognition and bioinformatics
- the alternative name 'Kohonen map'