Kohonen Maps, aka. Self-organizing maps (SOMs) are neural networks that visualize a high-dimensional feature space on a low-dimensional map. While SOMs are an excellent tool for data examination and exploration, they inherently cause a loss of detail. Visualizations of the underlying data do not integrate well and, therefore, fail to provide an overall picture. Consequently, we suggest SOMson, an interactive sonification of the underlying data, as a data augmentation technique. The sonification increases the amount of information provided simultaneously by the SOM.

In a first example, SOMson is introduced, sonifying a map with a four-dimensional feature space. Clicking on the map and moving the mouse, sonfies the all four features simultaneously. If you first want to train how to use the map, you can move the sliders to hear the influence of the single psycho-acoustic parameters. Furthermore, you can use the buttons above the map to visualize how certain features are distributed over the map. Further, clicking the "1D"-Button above the sliders limits the sonification to the selected features.

The extended example sonfies a seven-dimensional feature space. The basic functions are the same as in the first example, but you can also select which acoustical parameter represents which feature of the SOM using the modulation matrix. Linked combinations are highlighted in green. Clicking them a second time inverts the mapping, which is displayed in red.

The associated publication is currently under review and available as preprint.

Linke, S. & Ziemer, T.: SOMson -- Sonification of Multidimensional Data in Kohonen Maps. arXiv:2404.00016 [cs.HC], 2024

A further demonstration of how to interact with SOMson is provided as a YouTube playlist.