Hands-On Unsupervised Learning with Python

Self-organizing maps

A self-organizing map is a model that was proposed for the first time by Willshaw and Von Der Malsburg (in How Patterned Neural Connections Can Be Set Up by Self- Organization, Willshaw, D. J. and Von Der Malsburg, C., Proceedings of the Royal Society of London, B/194, N. 1117, 1976), with the goal of finding a way to describe different phenomena that happen in the brains of many animals. In fact, they observed that some areas of the brain can develop internally organized structures whose subcomponents are selectively receptive, with respect to specific input patterns (for example, some visual cortex areas are very responsive to vertical or horizontal bands). The central idea of an SOM can be synthesized by thinking about ...