Self-Organizing Maps (SOM) are unsupervised neural networks that map high-dimensional data by projecting it onto a low-dimensional, usually two-dimensional grid structure.
The training is done through competitive learning, where neurons compete for the input and the winner as well as its neighbors are adjusted.
This creates a topological mapping in which similar input patterns activate neighboring neurons.
SOMs assist in visualization, clustering, and dimensionality reduction of data and are used in fields such as image processing, bioinformatics, or market research.
The quality of the mapping depends on parameters such as learning rate, neighborhood function, and training duration.