模糊ISODATA聚类的BP神经网络及其应用

BP Neural Network Based on Fuzzy ISODATA Clustering and its Applications

提出了一种基于改进的模糊ISODATA聚类的BP神经网络算法,增强了其处理大样本的分类能力,可以很好地解决大样本情况引起的网络结构复杂、收敛性和泛化能力差等神经网络的固有问题。将其应用于消费者个人信用评估中,通过实验对比表明该算法精度较高,容错性好。

A new algorithm,called BP neural network based on improved fuzzy ISODATA clustering is proposed and the classification ability to deal with the big sample is enhanced. This new method can be used to solve some internal problems such as complex net structure,weak constringency and pan-ability.And applying it to personal credit assessment,the contrast experiment shows this algorithm has high precision and good performance.