A parallel hierarchical clustering algorithm based on PRAM model

An adaptive parallel algorithm for hierarchical clustering based on PRAM model was presented. Performing the data preprocessing depended on “90-10” rule to decrease the numbers of data set, performing the parallel algorithm for creating Euclid Minimum Spanning Trees on absolute graph, performing the algorithm for finding the disjoining strategies and non-collision memory, data set was clustered optimizedly. Data set was clustered on the conditions of non-collision memory, lowest-cost and weakest PRAM-EREW model. N data sets were clustered in O((λn)2/p) time (0.1≤λ≤0.3) performing this algorithm on p processors (1≤p≤n/log(n)). The parallel clustering algorithm based on PRAM model is an adaptive non-collision memory parallel hierarchical clustering algorithm. The calculating time will be greatly reduced after original inputing data are effectually preprocessed through improved preprocessing methods of this thesis.