基于神经网络模型的聚类分析技术研究

Study of clustering analysis technology based on neural netwoks model

聚类分析已成为数据挖掘,模式识别等应用领域研究中非常活跃的研究课题.在聚类分析方法中,基于神经网络的算法,由于考虑到“噪声”或异常数据,可以自动确定聚类个数,可以产生鲁棒的聚类方法,而竞争学习神经网络、SOFM神经网络方法是其中有代表性的方法,对其进行了分析研究,并给出了引入可变速度的训练算法。

Clustering analysis has been an active field in datamining and pattern recognition. Clustering analysis methods based model can know the numbers of the classes on the basis of analysing noisy data or abnormal data, robust clustering analysis methods can be acquired. Competitive learning NN and SOFM NN are representative methods, deeply analysis and study has been finished and an algorithm with variable speed feature has been given as well as.