摘要
针对传统Kohonen自组织特征映射(SOFM)神经网络模型结构需要预选指定的限制,特别在大的映射网络中寻找最佳匹配结点是很耗时的问题,我们采用一种新的动态增长树型自组织特征神经网络(GTS-SOFM),给出了实现聚类的具体算法,并且使用聚类密度来衡量聚类效果.对样本进行随机抽样,实验结果证实了算法的有效性.
One of the drawbacks of the SOFM is that the user must select the map size in advance, especially the time-consuming search for the best matching unit in large maps, A new Growing Tree-Structured Self-Organizing Maps （GTS-SOFM） is proposed and the specific algorithm to implement clustering is given, by using cluster density to Measure Cluster Quality. With the help of random sample technique, the experiment proves the efficiency of algorithm.
出处
《安徽工程科技学院学报（自然科学版）》
CAS
2007年第1期67-70,共4页
Journal of Anhui University of Technology and Science
