Self-Organizing Maps SOM is an unsupervised learning algorithm derived from competitive learning, which uses a proximity function to maintain the topological properties of the input space, usually represented by a low-dimensional discretization, in order to train the input space of samples.

SOM Typical Structure

The typical structure of SOM has two layers, namely the input layer and the output layer. The input layer simulates the retina that perceives external input information, and the output layer simulates the cerebral cortex that responds. The figure below is a schematic diagram of two SOM networks in 1D and 2D.

SOM and Self-Organizing Neural Networks

SOM is a self-organizing neural network. In addition to SOM, common self-organizing neural networks include Counter propagation network, Adaptive Resonance Theory network, etc.

Parent term: Self-organizing network

Synonyms: Dual Propagation Network, Adaptive Resonance Theory Network

References

【1】https://zh.wikipedia.org/wiki/Self-Organizing-Map

【2】https://blog.csdn.net/xbinworld/article/details/50826892

【3】https://www.cnblogs.com/sylvanas2012/p/5117056.html