This network architecture was created by the Finnish professor Teuvo Kohonen at the beginning of the 80s. It consists of one single layer neural network capable of providing a visualization of the data in one or two dimensions.
In this book, we are going to use Kohonen networks also as a basic competitive layer with no links between the neurons. In this case, we are going to consider it as zero dimension (0-D).
Theoretically, a Kohonen Network would be able to provide a 3-D (or even in more dimensions) representation of the data; however, in printed material such as this book, it is not practicable to show 3-D charts without overlapping some data. Thus in this book, we are going to deal only with 0-D, 1-D, and 2-D Kohonen networks.
Kohonen Self-Organizing Maps (SOMs), in addition to the traditional single layer competitive neural networks (in this book, the 0-D Kohonen network), add the concept of neighborhood neurons. A dimensional SOM takes into account the index...