An optimum ECG denoising with wavelet neural network

In the proposed paperECG denoising is achieved using wavelet neural network by approximating signal to a maximum possible accuracy. The feed forward back propagation neural network with ten neurons and two hidden layers is designed with conjugate gradient optimization.The library wavelet such as daubachies, symletetc are used as activation function for one hidden layer for ECG signal estimation.The performance achieved with db6 wavelet is found to be superior. Alsosignal to noise ratio and mean square error obtained with wavelet neural network are compared with soft thresholding the discrete wavelet coefficients of ECG signal by traditional methods.This shows the denoising achieved by estimating the signal using neural networks outperforms the others.