摘要
讨论了基于径向基函数 ( RBF)的概率神经网络的基本网络结构和网络的学习和运行过程 ,并且与 BP算法的径向基神经网络进行了对比 ,同时也测试了网络的容错能力 .结果表明 ,基于RBF的概率神经网络 ,学习速度大大提高 ,同时减小了 BP陷入局部极小的问题 ,有一定的抗噪声能力 .基于
An arrhythmia detection algorithm using radial basis function (RBF) based probabilistic neural network (PNN) was proposed. RBFNN has been widely applied in pattern recognition, but often with BP algorithm, spoiling its performance. This article discussed the configuration, learning and running of the network, compared it with BP algorithm based RBF and tested its tolerance to errors. The results of the experiments show that it avoids the problem of local minimum of BP algorithm and has capability in noise tolerance, while raises the learning rate. It has an efficient application in arrhythmia classification.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2000年第11期1475-1477,共3页
Journal of Shanghai Jiaotong University
关键词
概率神经网络
心律失常
径向基函数
心电信号
probabilistic neural network
arrhythmia
radial basis function
neural networks
