基于信息熵改进的 K-means 动态聚类算法

初始聚类中心及聚类过程产生的冗余信息是影响K-means算法聚类性能的主要因素,也是阻碍该算法性能提升的主要问题.因此,提出一个改进的K-means算法.改进算法通过采用信息熵对聚类对象进行赋权来修正聚类对象间的距离函数,并利用初始聚类的赋权函数选出质量较高的初始聚类中心点;然后,为算法的终止条件设定标准阈值来减少算法迭代次数,从而减少学习时间;最后,通过删除由信息动态变化而产生的冗余信息来减少动态聚类过程中的干扰,以使算法达到更准确更高效的聚类效果.实验结果表明,当数据样本数量较多时,相比于传统的K-means算法和其他改进的K-means算法,提出的算法在准确率和执行效率上都有较大提升.

Improved K-means dynamic clustering algorithm based on information entropy

Initial cluster centers and redundant information which is generated in clustering process will affect the clustering performance of K-means algorithm. In order to overcome the above mentioned shortcomings, a modified K-means algorithm is proposed. Firstly, it uses information entropy empowering the clustering objects to correct the distance function, and then employs empowerment function to select the optimal initial cluster centers. Subsequently, it decreases algorithm iterations to reduce learning time by setting the threshold value for termination condition of the algorithm. Finally,it reduces interference of dynamic clustering by removing redundant information from clustering process to make the proposed algorithm achieve more accurate and efficient clustering effect. The experimental results show that, when the data sample is larger, compared with the traditional K-means algorithm and other improved K-means algorithm, this improved K-means algorithm has large improvement in accuracy and efficiency.