k均值聚类中的EM思想.docVIP

k均值聚类中的EM思想 　　【摘 要】在无监督学习中，k均值聚类以其快速简单的特点得到了广泛的应用。EM算法是针对缺失数据的一种统计学习方法。然而，k均值和EM这两种不同领域的算法在思想上却有着一致的地方。本文分析了k均值中蕴含的EM思想，指出了k均值中样本隶属度更新和类中心更新与EM算法中的E步和M步的等价性。最后，利用R语言矩阵化运算的特点，介绍在如何在R语言中高效地实现k均值聚类算法。 　　【关键词】k均值；EM算法；聚类分析 　　【Abstract】In unsupervised learning， k-means clustering is widely applied in many fields due to the fact that it is very simple and fast. EM algorithm is a statistical learning approach for missing data. Although these two methods are applied in different areas， they are similar in terms of some ideas. The principle of EM implied in k-means clustering is analyzed in this