Showing 1–1 of 1 results for author: Koepke, H A

A General Hybrid Clustering Technique

Authors: Saeid Amiri, Bertrand Clarke, Jennifer Clarke, Hoyt A. Koepke

Abstract: Here, we propose a clustering technique for general clustering problems including those that have non-convex clusters. For a given desired number of clusters $K$, we use three stages to find a clustering. The first stage uses a hybrid clustering technique to produce a series of clusterings of various sizes (randomly selected). They key steps are to find a $K$-means clustering using $K_\ell$ cluste… ▽ More Here, we propose a clustering technique for general clustering problems including those that have non-convex clusters. For a given desired number of clusters $K$, we use three stages to find a clustering. The first stage uses a hybrid clustering technique to produce a series of clusterings of various sizes (randomly selected). They key steps are to find a $K$-means clustering using $K_\ell$ clusters where $K_\ell \gg K$ and then joins these small clusters by using single linkage clustering. The second stage stabilizes the result of stage one by reclustering via the `membership matrix' under Hamming distance to generate a dendrogram. The third stage is to cut the dendrogram to get $K^*$ clusters where $K^* \geq K$ and then prune back to $K$ to give a final clustering. A variant on our technique also gives a reasonable estimate for $K_T$, the true number of clusters. We provide a series of arguments to justify the steps in the stages of our methods and we provide numerous examples involving real and simulated data to compare our technique with other related techniques. △ Less

Submitted 5 March, 2015; v1 submitted 3 March, 2015; originally announced March 2015.