A Parallel Tabu Search Algorithm with Solution Space Partition for Cohesive Clustering Problems

Clustering analysis plays an important role in a wide range of fields including data mining, pattern recognition, machine learning and many other areas. In this paper, we present a parallel tabu search algorithm for clustering problems. A permanent tabu list is proposed to partition the solution space for parallelization. Moreover, this permanent tabu list can also reduce the neighborhood space and constrain the election of candidates. The proposed approach is evaluated by clustering some specific dataset. And experimental results and speedups obtained show the efficiency of the parallel algorithm.