Parallelization of the K-means fast learning artificial neural network

The paper presents a parallelization study of an improved K-means fast learning artificial neural network (K-FLANN) within the parallel virtual machine (PVM) environment. The study discusses the improvements made on the K-FLANN II algorithm, which eventually lead to a consistent set of cluster centroids, regardless of the data presentation sequence DPS. To further improve clustering efficiency, a form of hierarchical clustering is explored, leading to the parallel-distributed implementation of the K-FLANN. Results of the investigation are presented along with a discussion of the fundamental behavior of the parallel network.