A Generative Adversarial Approach to ECG Synthesis And Denoising

Abstract:

Generative Adversarial Networks (GAN) are known to produce synthetic data that are difficult to discern from real ones by humans. In this paper, we present an approach to use GAN to produce realistically looking ECG signals. We utilize them to train and evaluate a denoising autoencoder that achieves state-ofthe- art filtering quality for ECG signals. It is demonstrated that generated data improves the denoiser performance compared to the model trained on real data only. We also investigate an effect of transfer learning by reusing trained discriminator network for the denoising model.