【作者】 蒋余庆； 巩敦卫； 周勇； 张勇；

【Author】 JIANG Yu-qing1,GONG Dun-wei1,ZHOU Yong2,ZHANG Yong1(1.School of Information and Electronic Engineering,China University of Mining and Technology,Xuzhou 21008,China;2.School of Computer Science and Technology,China University of Mining and Technology,Xuzhou 21008,China)

【机构】 中国矿业大学信息与电气工程学院； 中国矿业大学计算机科学与技术学院；

【摘要】 针对K-均值聚类算法中初始聚类数目难以确定,对初始参数敏感等问题,提出一种快速遗传聚类算法,算法中采用可变长实数表示聚类中心,并设计新的交叉和变异算子,以及采用DB-Index准则作为聚类的目标函数。实验结果表明:该算法不但具有较优的聚类性能,而且收敛速度也大大提高。更多还原

【Abstract】 In order to solve problems of K-means algorithms,such as difficult to determine the initial number of clusters and sensitive to the initial values of some pertinent parameters,a fast genetic clustering algorithm was proposed,in which clustering centers were coded by real number with variable length,novel crossover and mutation operators were designed,and DB-index criterion was adopted as the objective function of clustering.The experimental results show that the algorithm proposed has not only good performance in clustering but also fast speed of convergence.更多还原