Solving mean-shift clustering using MapReduce Hadoop

Paper presents results of a practical experiment that was conducted in order to pursuit main objective - design and implement novel iterative MapReduce framework based on Hadoop technology. To study the effects of implementation parallel scientific applications, we deployed. Despite the fact that they did not show better performance than the same algorithm implemented in MPI, we conducted experiments to solve problems iterative computations using the Hadoop architecture. As a result of the experiments, the fail-safe behavior of Hadoop technology was revealed in the solutions of complex hadacies.