使用BP网络改进K-means聚类效果

Using BP Neural Network to Improve K-mean' Result

K-means 算法中的 k 值的确定和初始聚类中心的选择严重影响聚类效果。针对这一问题,本文提出使用 BP神经网络改进 K-means 聚类效果的方法。通过对聚类结果进行反复训练,调整聚类数,K-means 的聚类效果得到改善。采用人工数据和实际商业数据的实验证明该方法能有效地改善传统的聚类效果。

The value of K and the selection of initial centers heavily affect the result of K-means algorithm. Aiming at this problem, this paper puts forward a method that uses BP neural network to improve K-means＇ result. Training clustering result by BP network, the effect of K-means can be improved. Experiments using artificial data and actual business data testify the validity of this method. It can improve the traditional K-means effect well.