https://hgpu.org/?p=14410
A Parallel Implementation of the Self Organising Map using OpenCL