基于自组织特征映射神经网络的局部矢量量化算法  

A LOCAL LEARNING VECTOR QUANTIZATION ALGORITHM BASED ON SELF-ORGANIZING FEATURE MAPPING NEURAL NETWORKS

作　　者：缪青[1] 高大启[1] 

机构地区：[1]华东理工大学计算机系,上海200237

出　　处：《计算机应用与软件》2006年第7期108-109,112,共3页Computer Applications and Software

摘　　要：提出了一种基于自组织特征映射神经网络的局部矢量量化算法(Local vector quantization algorithm based on Self-O rgan i-zing Feature M app ing neural networks,LSOFM),LSOFM算法是对SOFM算法的一种改进,它将隶属关系引入到参考点权值的修改中,自组织特征映射神经网络的领域大小的确定依赖于训练矢量与参考点之间的隶属关系。This paper presents a kind of local learning vector quantization algorithm which bases on self-organizing feature mapping neural networks （LSOFM）. Improving on SOFM, The LSOFM algorithm introduces subjection relation into the process of prototype design. The domain used in self-organizlng feature mapping neural networks depends on the subjection relation between the learning vector and the prototype.

关 键 词：LVQ SOFM LSOFM 隶属度 领域 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP391.41[自动化与计算机技术—计算机科学与技术]