摘要
文章对BP神经网络中较为常用的梯度函数进行比较分析。对这几个梯度函数在UCI数据集上的分类以及对非线性函数的回归预测进行matlab仿真,并分析比较分类准确率和非线性函数回归预测精确度。最后给出应用BP神经网络进行仿真时梯度函数的选择方法。
Comparative analysis of frequently-used gradient functions of BP artificial neural networks is listed in this paper.It performs simulation of classification on UCI dataset and regressive prediction on nonlinear functions with these gradient functions.Accuracy rate of classification and precision of prediction are analyzed after simulation.Finally,the way of choosing gradient training function of BP is given.
出处
《福建教育学院学报》
2011年第1期120-122,共3页
Journal of Fujian Institute of Education
关键词
BP网络
梯度函数
回归预测
分类
BP
Gradient Function
regressive prediction
classification
