The ordering of CH function is first discussed. The generalized P ordering CH function is derived. And then the general presentation of CH function with different ordering is presented by means of the multiplication o...The ordering of CH function is first discussed. The generalized P ordering CH function is derived. And then the general presentation of CH function with different ordering is presented by means of the multiplication of generalized Radamache functions. Moreover, the construction property of CH function is studied in detail, and the copy theory of CH function is developed. Finally, the generalized bridge function is constructed on the basis of copy theory.展开更多
针对传统的K-medoids聚类算法在聚类时需要随机选择初始类中心且指定聚类数目K,及聚类结果不稳定的问题,提出了一种优化初始类中心的自适应K-medoids算法(adaptive K-medoids algorithm for optimizing initial class centers,CH_KD)....针对传统的K-medoids聚类算法在聚类时需要随机选择初始类中心且指定聚类数目K,及聚类结果不稳定的问题,提出了一种优化初始类中心的自适应K-medoids算法(adaptive K-medoids algorithm for optimizing initial class centers,CH_KD).其思想是定义了特征重要度,以此筛选出每一簇中最优的代表特征,组成特征子集,并重点研究了传统划分算法的自适应优化与改进.首先,利用特征标准差定义特征区分度,选择出区分度强的特征.其次,利用皮尔逊相关系数度量特征簇中每个特征的冗余度,选择出冗余度低的特征.最后,将特征区分度与特征冗余度之积作为特征重要度,以此筛选出每一簇中最优的代表特征,组成特征子集.实验将所提算法与其他聚类算法在14个UCI数据集上进行对比,结果验证了CH_KD算法的有效性与优势.展开更多
文摘The ordering of CH function is first discussed. The generalized P ordering CH function is derived. And then the general presentation of CH function with different ordering is presented by means of the multiplication of generalized Radamache functions. Moreover, the construction property of CH function is studied in detail, and the copy theory of CH function is developed. Finally, the generalized bridge function is constructed on the basis of copy theory.
文摘针对传统的K-medoids聚类算法在聚类时需要随机选择初始类中心且指定聚类数目K,及聚类结果不稳定的问题,提出了一种优化初始类中心的自适应K-medoids算法(adaptive K-medoids algorithm for optimizing initial class centers,CH_KD).其思想是定义了特征重要度,以此筛选出每一簇中最优的代表特征,组成特征子集,并重点研究了传统划分算法的自适应优化与改进.首先,利用特征标准差定义特征区分度,选择出区分度强的特征.其次,利用皮尔逊相关系数度量特征簇中每个特征的冗余度,选择出冗余度低的特征.最后,将特征区分度与特征冗余度之积作为特征重要度,以此筛选出每一簇中最优的代表特征,组成特征子集.实验将所提算法与其他聚类算法在14个UCI数据集上进行对比,结果验证了CH_KD算法的有效性与优势.
