K-means 聚类算法是近年来数据挖掘学科的一个研究热点和重点,该算法是基于划分的聚类分析算法。目前这种算法在聚类分析中得到了广泛应用。本文将介绍K-means聚类算法的主要思想,及其优缺点。针对该算法经常陷入局部最优,以及对孤立点敏感等缺点,提出了一种基于模拟退火算法的方法对其进行优化,可以有效地防止该算法陷入局部最优的情况。% K-means clustering algorithm, which is a division-based clustering and analysis algorithm, has become a hotspot of data-mining subject in recent years. Now this algorithm has been widely applied in the clustering analysis. In this article, we introduced the main idea and advantages /disadvantages of the K-means clustering algorithm. Aiming to the defects of this algorithm such as local optimum and sensitive to isolated points, we suggested a simulation-based annealing algorithm to optimize it so as to prevent the algorithm from experiencing local optimum efficiently.
展开▼