Cluster detection of special datasets using the PYRAMID algorithm

Clustering is the art of discovering patterns in large data sets. With the constantly growing computational demands of modern applications, such data sets have grown tremendously in size and complexity, imposing further challenges on clustering algorithms. This includes outlier handling, detection of arbitrary shaped clusters, processing speed, and dependence on user-supplied parameters. The latest decade has witnessed many studies that have addressed one or more of these challenges. One of these is PYRAMID, or parallel hybrid clustering using genetic programming and multi-objective fitness with density, which we introduced in a previous research. PYRAMID is an algorithm that addresses some of the above challenges by employing a combination of data parallelism, a form of genetic programming, and a multi-objective density-based fitness function in the context of clustering. This study adds to our previous research by analyzing some of the detection characteristics of PYRAMID with respect to a set of challenging datasets and drawing conclusions as well as future directions.