Nonparametric Nearest Neighbor Descent Clustering based on Delaunay Triangulation

In our physically-inspired clustering algorithm and the series after it, the only free parameter is involved in computing the potential value of each point. In this work, we propose a nonparametric process to compute potential values based on the Delaunay Triangulation or Voronoi. This can make all our previous clustering algorithms nonparametric. We demonstrate the power of these nonparametric clustering algorithms by several clustering examples.