A Study of ECG Characteristics by Using Wavelet and Neural Networks

ECG consists of various waveforms of electric signals of heat. Machine Learning methods such as the MLP classification have proven to perform well in ECG classification. In this study, preprocessing was performed through wavelet transform, and in classification several characteristics were evaluated using BP algorithm that applied generalized delta rules to MLP. In order to decide wavelet generating function that can remove baseline by minimizing the distortion of raw signals, this study removed baseline by applying various wavelet generating functions. To evaluate the results above according to the learning method, learning iteration and learning rate of neural networks, various experiments were conducted.