Unsupervised learning: the Dog Rabbit strategy
P. McKenzie, Michael D. Alder · 2002
We describe a method of untrained learning called the Dog Rabbit strategy that finds cluster centers in data sets, and compares it to the well known k-means clustering algorithm on data with Gaussian distributions. The Dog Rabbit strategy is an iterative procedure that uses a dynamic process to move k points, or neurons, to positions near the centres of clusters in a data set.>