【摘要】 启发式聚类算法的搜索空间中布满了局部极小值"陷阱",从而使得算法容易过早收敛而无法获得高质量聚类结果。文章给出了一种噪声启发式聚类算法NHCA(Noising Heuristic Clustering Algorithm),该算法在搜索空间中增加一组由强至弱的噪声来扩大启发式搜索的局部范围,以保持搜索空间的多样性,达到避免局部极小值影响和提高聚类质量的目的。大量实验结果表明,噪声法对提高启发式聚类算法质量是十分有效的。更多还原

【Abstract】 The heuristic clustering algorithm easily converges to the local optimal result,for there are lots of local minimum "stucks" in its search space,so the heuristic clustering algorithm can not give high quality results.A new clustering algorithm,which is called Noising Heuristics Clustering Algorithm(NHCA),is proposed in this paper.The main idea is that a series of noises from strong to weak is added to the heuristic clustering algorithm search space to expand its local search scope.Thus the diversity of the search space is retained,the influence of the local minimum is avoided,and high quality clustering results can be gotten.Experiments show that the noising method is helpful in improving the cluster quality of the heuristic clustering algorithm.更多还原