一种基于变邻域搜索的启发式聚类算法

A Heuristic Clustering Algorithm Based on Variable Neighborhood Search

聚类是数据挖掘的核心问题,已经成为新的研究热点。从组合优化角度描述的聚类问题是NP-难解的,因此,研究者们通常采用启发式搜索对该类问题进行求解。但是搜索空间的复杂性使得启发式聚类算法存在易受初始解等因素影响的缺点。采用变邻域搜索技术引导启发式聚类算法在解空间中搜索,给出了一种新的启发式聚类算法。该算法很好地避免了启发式聚类算法初始解敏感的问题,有效地提高了聚类质量。

Clustering is a kernel problem in data mining area which has become a hot research topic.The clustering problem which was described by combinatorial optimization is NP-hard,so the researchers to solve it by using heuristic search method.For the complexity of search space,thereby,heuristic clustering algorithms are easy be effected by initial result.In this paper,we propose a novel heuristic clustering algorithm which was guided by variable neighborhood search.Experimental results show that the proposed algorithm has ability to deal with initialization problem and could improve the quality of clustering result.