A Wavelet Neural Network Algorithm of EEG Signals Data
Compression and Spikes Recognition
Zhang Yongsheng Liu Aiping Yu Ke
Department of Computer Science, Wuhan University of Technology, Wuhan 430070
Departemnt of Computer, Luoyang Technology College, Luoyang 471003
Optical Fiber Sensing Research Center, Wuhan University of Technology, Wuhan 430070
Abstract A novel method of EEG signals compression representation and epileptiform spikes recognition based on wavelet neural network and its algorithm is presented. The wavelet network not only can compress data effectively but also can recover original signal. In addition, the characters of the spikes and the spike-slow rhythm are auto-detected from the time-frequency isoline of EEG signal. This method is well worth using in the field of the electrophysiological signal processing and time-frequency analyzing.
Key words Electroncephalograph Wavelet neural network Data compression Signal processing Epilepsy
1 基本原理
1.1 数据压缩表达网络
小波变换实质是一种不同参数空间之间通过小波基进行的积分变换［4］。小波基是此变换的内核，选择的不同对变换起着关键的作用。小波网络则是基于小波分析而构造的新的神经网络模型，其思想是用非线性小波基取代了通常的非性Sigmoid函数，其信号表述是通过将所选取的小波基进行线性叠加来实现的，信号s(t)可用小波基进行如下拟合：
[1] 2 3 4 5 6 7 下一页
