Design of a Structured 3D SOM as a Music Archive

Abstract

A structured 3D SOM is an extension of a Self-Organizing Map from 2D to 3D where a structure has been built into the design of the 3D map. The 3D SOM is a 3x3x3 cube, with a distinct core cube in the center, and 26 exterior cubes around the center. The structured SOM mainly uses the 8 corner cubes among the 26 exterior cubes. Used to build a music archive, the SOM learning algorithm is modified to include a four-step learning and labeling phase. The first phase is meant only to position the music files in their general locations within the core cube. The second phase is meant to position the music files in their respective corner cubes. The third phase is meant to do a fine adjustment of the weight vectors in the core cube. The fourth phase is the labeling of the map and the association of music files to specific nodes in the map. Through the embedded structure of the 3D SOM, a precise measure is developed to measure the quality of the resulting trained SOM (in this case, the music archive), as well as the quality of the different categories/genres of music albums based on a novel measure of the attraction index and the fidelity of music files to their respective music genres.