计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2014年
22期
149-153
,共5页
最近邻%自适应聚类%最小生成树%聚类分析
最近鄰%自適應聚類%最小生成樹%聚類分析
최근린%자괄응취류%최소생성수%취류분석
nearest neighbor%adaptive clustering%minimum spanning tree%clustering analysis
针对传统最小生成树聚类算法需要事先知道聚类数目和使用静态全局分类依据,导致聚类密度相差较大时,算法有效性下降,计算复杂度大等问题,提出一种改进的最小生成树自适应分层聚类算法,根据最近邻关系,自动为每个聚类簇设定独立的阈值,使之适应分布密度相差较大的情况,并能自动确定聚类数目。实验表明,算法具有较好的性能,尤其对数据密度分布不均匀的情况也能得到较好的聚类结果。
針對傳統最小生成樹聚類算法需要事先知道聚類數目和使用靜態全跼分類依據,導緻聚類密度相差較大時,算法有效性下降,計算複雜度大等問題,提齣一種改進的最小生成樹自適應分層聚類算法,根據最近鄰關繫,自動為每箇聚類簇設定獨立的閾值,使之適應分佈密度相差較大的情況,併能自動確定聚類數目。實驗錶明,算法具有較好的性能,尤其對數據密度分佈不均勻的情況也能得到較好的聚類結果。
침대전통최소생성수취류산법수요사선지도취류수목화사용정태전국분류의거,도치취류밀도상차교대시,산법유효성하강,계산복잡도대등문제,제출일충개진적최소생성수자괄응분층취류산법,근거최근린관계,자동위매개취류족설정독립적역치,사지괄응분포밀도상차교대적정황,병능자동학정취류수목。실험표명,산법구유교호적성능,우기대수거밀도분포불균균적정황야능득도교호적취류결과。
Classical clustering algorithm based on the minimum spanning tree often needs to know the number of clusters beforehand and use static global threshold to cluster, which leads to the performance of the algorithm low and the compu-tation complex for the uneven distributed data. An improved adaptive hierarchical clustering algorithm based on minimum spanning tree is proposed, which automatically generates different thresholds for every cluster to adapt for the uneven dis-tributed data according to the nearest neighbor relationship and adaptively determines the number of clusters. Experiments demonstrate that this algorithm has good performance, especially could cluster effectively for the uneven distributed data.