摘要
原始的k-means算法[4]是从样本点的集合中随机选取K个中心,这种选取具有盲目性和随意性,它在很大程度上决定了算法的有效性.为消除选取初始中心的盲目性,应充分利用已有数据样本点的信息.采取对数据进行预处理的方式来选取初始中心.实验证明新的初始点的选取不仅提高了算法的计算效率,也提高了算法最终确定的聚类的精度.
Original k-means clustering algorithm is the means that selects K centers randomly from the data sample cluster .This selection is blind and random, and to a certain extent the validity of algorithm lies on the selection. In order to avoid the blindness of selection, we should make full use of the information of existing data sample dot. We make pre-treatment of the data to choose the initial center. The experiment improves not only the calculation efficiency of algorithm, but also the precision of ultimate clustering.
出处
《西南民族大学学报（自然科学版）》
CAS
2009年第1期198-200,共3页
Journal of Southwest Minzu University(Natural Science Edition)
