Self Organizing Map (SOM)

Self Organizing Map (SOM)

In 2014, the need for projection from multi-dimensional spaces toward 2D or 3D appears. From the 12 audio indicators provided by Echonest (now within spotify) to 2 dimensions or else 1 dimension. We had already used PCA (principal Component Analysis) but Self Organizing Map (SOM) and their algorithms were fascinating.

Therefore, I coded a generic java package.

The java class needs only a distance function over a set of points.

I set the topology only through the distance function. It specified the change of learning factors over iterations using a linear function and key points.

The following illustrations show various kinds of use.

A video shows the package in use, there.