Density Clustering with Basin Spanning Trees

Sören Hader, Fred Hamprecht

Aim

Basin spanning trees are an efficient method of clustering multivariate data according to their density. The root of each tree corresponds to a local density maximum. The algorithm allows the creation of arbitrary cluster shapes.

Methods

  • density estimation
  • neighborhood graphs like Delaunay triangulation

Examples

Publications

  • "Efficient density clustering using basin spanning trees" S. Hader, F. A. Hamprecht; Proceedings of the GfKl 2002, Springer (2003)