摘要
为了解决聚类分析中聚类数的确定问题,在SOFM神经网络的基础上,从聚类准则出发,通过试验对聚类准则的曲线特征进行了详细的分析和论证,设计出一种结构自适应的聚类神经网络,该网络能自动确定最佳的聚类数,并提出了一种减少计算量的改进算法。
Ascertain a problem to gathering the kind number's in clustering analysis having been in progress studying. On SOFM neural networks basis ,the criterion sets off from clustering,analyze and expound and prove detailed by testing that the characteristic has been in progress to clustering criterion curve, design out one kind of certainly fit in with clustering of structure neural networks, be a network's turn to be able to have ascertained the optimum clustering number voluntarily, and have suggested that one kind cuts down the algorithm calculating the amounts improvement.
出处
《世界科技研究与发展》
CSCD
2009年第6期1053-1054,共2页
World Sci-Tech R&D
关键词
神经网络
聚类分析
聚类准则
聚类神经网络
neural network
clustering analysis
clustering criterion
clustering neural networks