摘要
采用Ｄｅｌｔａ规则和Ｄｅｌｔａ－Ｂａｒ－Ｄｅｌｔａ规则的神经网络，用于根据头发和血清样品中金属含量对正常人与癌症患者的分类判别研究，对有关参数的影响作了讨论．结果表明，采用Ｄｅｌｔａ－Ｂａｒ－Ｄｅｌｔａ规则的神经网络训练速度明显快于Ｄｅｌｔａ规则的神经网络．
The neural networks with common Delta and Delta Bar Delta rules were used for classification of normal persons and lung cancer patients based on the metal contents in hair and serum samples. The effects of some parameters including learning rate, momentum, number of neurons in hidden layer were discussed. Results showed that the training process with Delta Bar Delta rule was faster than the one with common Delta rule.
出处
《分析科学学报》
CAS
CSCD
1998年第3期183-186,共4页
Journal of Analytical Science
基金
国家自然科学基金
国家教委优秀青年教师基金
关键词
神经网络
癌
分类判别
肺癌
Delta规则
Artificial neural network, Chemometrics, Cancer, Classification
