A Survey on Partitioning and Parallel Partitioning Clustering Algorithms

Learning is the process of generating useful information from a huge volume of data. Learning can be classified as supervised learning and unsupervised learning. Clustering is a kind of unsupervised learning. A pattern representing a common behaviour or characteristics that exist among each item can be generated. This paper gives an overview of partition clustering and parallel implementation of this clustering algorithm. It describes about the general working behaviour, the methodologies followed on these approaches and the parameters which affects the performance of these algorithms. Clustering is grouping input data sets into subsets, called 'clusters' within which the elements are somewhat similar. In general, clustering is an unsupervised learning task as very little or no prior knowledge is given except the input data sets. The tasks have been used in many fields and therefore various clustering algorithms have been developed. Clustering task is, however, computationally expensive as many of the algorithms require iterative or recursive procedures and most of real-life data is high dimensional. Therefore, the parallelization of clustering algorithms is inevitable, and various parallel clustering algorithms have been implemented and applied to many applications. In this paper, we review a partition clustering algorithms and their parallel versions as well. Although the parallel clustering algorithms have been used for many applications, the clustering tasks are applied as pre processing steps for parallelization of other algorithms too.