How it works...

Vector quantization is an algorithm used for signal compression, image coding, and speech. We use geometric criteria (the Euclidean distance) to find clusters. It is, therefore, an example of unsupervised training. It is a technique that allows the modeling of probability density functions through the distribution of prototype vectors. Vector quantization divides a large set of points (vectors) into clusters by using a similar number of points closer to them. Each cluster is illustrated by its centroid point (as in k-means).