1. Infinite training. Each data that feed to the algorithm will automatically consider as the training
examples. The VB program above is on this type.
2. Finite training. After the training is considered as finished (after it gives about the correct place of
mean). We start to make the algorithm to work by classifying the cluster of new points. This is done
simply by assign the point to the nearest centroid without recalculate the new centroid. Thus after the
training finished, the centroid are fixed points.