基于主成分的SFLABP网络模型在帕金森氏病分类中的应用

Based-on PCA of SFLABP Neural Network Model in Application of Parkinson Disease＇ s Classification

针对BP神经网络学习效率低、容易陷入局部最优等缺点,提出了一种基于主成分分析的混合蛙跳算法（Shuffle FrogLeaping Algorithm）优化的BP神经网络模型。使用主成分分析法对高维数据进行特征提取,作为网络输入;采用混合蛙跳算法优化BP神经网络的权系数和阈值,构建基于混合蛙跳算法神经网络的帕金森病分类模型。最后,以UCI中Parkinson数据为例,实验表明,新模型优于传统的BP网络。

For the shortcomings of BP neural network which is low learning efficiency and is easy to trap into local optimum, ac- cording to these problems, a new BP neural network model optimized by Shuffle Frog Leaping Algorithm based on Principal Component Analysis is proposed. Using Principle Component Analysis to extract the features of high dimensional data, the input variables; the bias of BP neural network are optimized by Shuffle Frog Leaping Algorithm and then build the classification model of Parkinson＇s disease based on SFLABP neural network. At last, taking the data of Parkinson from UCI for example, the experi- ment result demonstrates the new model is better than the traditional BP neural network.