张亮, 蓝要武, 韩英, 张正行, 安登魁. 人工神经网络用于中药材雷公藤和昆明山海棠的分类识别研究J. 药学学报, 1995, 30(2): 127-132.
引用本文: 张亮, 蓝要武, 韩英, 张正行, 安登魁. 人工神经网络用于中药材雷公藤和昆明山海棠的分类识别研究J. 药学学报, 1995, 30(2): 127-132.
L Zhang, YW Lan, Y Han, ZX Zhang , DK An, . CLASSIFICATION AND RECOGNITION OF TRIPTERYGIUM WILFORDII AND T.HYPOGLAUCUM BY ARTIF101AL NEUKAL NETWORKSJ. Acta Pharmaceutica Sinica, 1995, 30(2): 127-132.
Citation: L Zhang, YW Lan, Y Han, ZX Zhang , DK An, . CLASSIFICATION AND RECOGNITION OF TRIPTERYGIUM WILFORDII AND T.HYPOGLAUCUM BY ARTIF101AL NEUKAL NETWORKSJ. Acta Pharmaceutica Sinica, 1995, 30(2): 127-132.

人工神经网络用于中药材雷公藤和昆明山海棠的分类识别研究

CLASSIFICATION AND RECOGNITION OF TRIPTERYGIUM WILFORDII AND T.HYPOGLAUCUM BY ARTIF101AL NEUKAL NETWORKS

  • 摘要: 应用误差反向传播学习算法(BP算法)对中药材雷公藤和昆明山海棠浸出物的红外光谱进行分类识别。网络为3层结构,输入节点为9个,隐层节点为21个,输出节点为1。分类结果与XIM-CA法基本一致。此外,本文还考察了网络参数间的相互关系。

     

    Abstract: A BASIC program for the simulation of artificial neural networks was implementedon a 386SX. For classification and recognition of infrared(IR) spectra of extracts by means of aneural network, a back propagation model with one hidden layer and a sigmoid transfer function hasbeen proved to be available. The nine features selected by the Shannon information content were usedas input elements. The hidden layer contains 21 nodes and the output layer contains one node.

     

/

返回文章
返回