18 Self-Organizing Maps
The k-means algorithm does a fairly good job of organizing data, but as seen in the previous chapter it does not always provide the best results. A more powerful solution is the self-organizing map (SOM), a data organization algorithm that creates a mapping space that can be used as either a clustering tool or an associative memory (Kohonen 1982, 1990). This chapter will review the construction of the SOM with a simple application.
18.1 SOM Theory
The SOM is a set of vectors arranged in a lattice. The elements of the vectors are altered through training so that the boundaries are created within the lattice. These boundaries define the cluster regions.
The process begins with a set of data vectors that all have the length ...
Get Python for Bioinformatics now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
