摘要
采用主成分分析法对样本数据集进行预处理,将得到的新样本数据集输入神经网络,构建F/10家族木聚糖酶氨基酸组成和最适温度的主成分分析神经网络(PCANN)模型.结果表明,当学习速率为0.07、动态参数为0.8、Sigmoid参数为0.96,隐含层结点数为5时,模型对温度拟合的平均绝对百分比误差为4.97%,绝对误差为3.03℃.同时,方法具有良好的预测效果,预测的平均绝对百分比误差为4.68%,平均绝对误差为3.55℃.
The principal component analysis was first applied to the data processing in training sets, and then the obtained new principal components were used as input parameters of BP neural networks. A prediction model for optimum temperature of xylanases in F/10 family was established based on uniform design. When the learning rate, momentum parameter, Sigmoid parameter and the neuron numbers of the hidden layer was 0. 07,0. 8,0. 96 and 5, respectively, the eaiculated temperatures fitted the reported optimum temperatures very well. The mean absolute percent error was 4. 97%. At the same time, the predicted temperatures fitted the reported optimum temperatures well and the mean absolute error was 3.55 ℃.It was superior in fittings and predictions compared to the reported model based on stepwise regression,
出处
《华侨大学学报(自然科学版)》
CAS
北大核心
2007年第1期55-58,共4页
Journal of Huaqiao University(Natural Science)
基金
国务院侨务办公室科研基金资助项目(05Q0018)
关键词
主成分分析
BP神经网络
木聚糖酶
最适温度
虚拟筛选
principal component analysis
BP neural networks
xylanase
optimum temperature
virtual screening