基于自组织特征映射神经网络的储粮害虫分类方法研究  

Study on Stored-grain Pests Classification Based on Self-organizing Feature Map Neural Network

作　　者：梁斌梅[1] 

机构地区：[1]广西大学数学与信息科学学院,广西南宁530004

出　　处：《安徽农业科学》2009年第32期16156-16158,共3页Journal of Anhui Agricultural Sciences

基　　金：广西教育厅科研项目([2006]026)

摘　　要：提出了基于自组织特征映射(SOM)神经网络的害虫分类方法。该方法能将任意维输入模式在输出层映射成一维或二维离散图形,并保持其拓扑结构不变,而且无需监督,可实现对输入模式自动分类。分析了SOM网络基本工作原理,并将之用于害虫分类模型的建立中。结果表明,该方法能有效地对害虫进行分类,比BP神经网络分类精确度高、分类结果的可解释性更好。A classification method based on self-organizing feature map （SOM） neural network was proposed. The high-dimensional input space could be projected into one-dimension or two-dimension discrete space with the method. The classification process didn＇ t need supervision and could make input space to be classified automatically. The working principle of SOM network was analyzed, and the SOM network was used to build the stored-grain pests classification model. Results showed that this method could classify pests effectively, and showed higher classification accuracy and better understandability than BP neural network.

关 键 词：自组织特征映射 神经网络 储粮害虫 分类 

分 类 号：S126[农业科学—农业基础科学]