A method for the analysis of Gaussian-like clusters

摘要

In this paper we are concerned with the analysis of Gaussian-like point clusters in an n-dimensional space. The method described uses the directed fuzzy graph obtained from the points by linking every point to its neighbors and assigning a measure of the local point density to each node of the graph. Next, the tops of the fuzzy graph are found and a subset of them selected as cluster centers. The rest of the points are then classified based on the selected tops.