摘要
介绍了对向传播神经网络的原理、算法。以酚类化合物的8个结构特征参数为输入,用对向传播神经网络对酚类化合物的毒性进行模式分类识别。结果表明,对向传播神经网络具有较强的模型拟合能力和泛化能力。网络对36个训练样本和8个预测样本的毒性类型都能进行准确识别,是一种有效的模式分类识别方法。
This paper introduces the principle and algorithm of counter propagation network.Which using eight structural characteristic parameters of phenolic compounds as inputs to classify and recognize the toxicity of phenolic compound by counter propagation network.The results indicate that the counter propagation network has strong model fitting and generalization abilities.The network can accurately identify the toxicity types of 36 training samples and 8 prediction samples.It is an effective pattern classification recognition method.
作者
申明金
Shen Mingjin(School of Pharmacy,North Sichuan Medical College,Nanchong 637000,China)
出处
《山东化工》
CAS
2024年第5期221-224,共4页
Shandong Chemical Industry
关键词
对向传播神经网络
酚类化合物
毒性
模式识别
counter propagation network
phenolic compound
toxicity
pattern recognition
