Coring Method for Clustering a Graph
This is a new clustering method for weighted/unweighted graphs. It identifies and expands cores of clusters. Its time complexity for a weighted graph G=(V,E) is O(|E| + |V| log |V|). Results are surprisingly good!
The cores!
Download my slides here.
Experiments with image segmentation
Examples
An 1D example:
Dataset of 110 random points generated by 2 Normal distributions:
60 points from N(0, 0.8) and 50 points from N(2, 0.3).
2D examples
