Self-Organizing Maps

Self-Organizing Maps (SOMs) are a type of artificial neural network used for unsupervised learning and data visualization, developed by Teuvo Kohonen. They reduce high-dimensional data into a low-dimensional (typically 2D) grid of nodes while preserving topological relationships, making them useful for clustering, pattern recognition, and exploratory data analysis. SOMs learn through competitive learning, where nodes adjust their weights to represent input data patterns.

Also known as: SOM, Kohonen Maps, Kohonen Networks, Self-Organizing Feature Maps, SOFM

Why learn Self-Organizing Maps?

Developers should learn SOMs when working with high-dimensional datasets where visualizing or clustering complex patterns is needed, such as in customer segmentation, image analysis, or anomaly detection. They are particularly valuable in fields like bioinformatics, finance, and marketing for discovering hidden structures in data without labeled examples, offering an intuitive way to interpret results through a 2D map.

Learning Resources

• tutorial: Kohonen's Self-Organizing Maps
• tutorial: Self-Organizing Maps Explained
• video: SOM Tutorial on YouTube
• docs: Scikit-learn SOM Implementation