Denoising Autoencoder for Biosignals: Denoising Autoencoder for Biosignals
Authors
Jensen, Simon Anielski Barsøe ; Rasmussen, Rasmus Hjelm
Term
4. term
Education
Publication year
2021
Submitted on
2021-06-17
Pages
16
Abstract
Biosignals are measurements of the body’s activity, such as heart rhythm (ECG), brain activity (EEG), and muscle activity (EMG). These data often contain noise from movement, equipment, or the environment, which can hide important patterns. A denoising autoencoder is a type of neural network that learns to reconstruct a clean version of a signal from its noisy input. This thesis examines the use of denoising autoencoders to clean biosignals and outlines key considerations, such as preserving relevant signal details and avoiding the removal of clinically important information. The aim is to make biosignals clearer and more useful for subsequent analysis.
Biosignaler er målinger af kroppens aktivitet, for eksempel hjerterytme (ECG), hjerneaktivitet (EEG) og muskelaktivitet (EMG). Disse data indeholder ofte støj fra bevægelse, udstyr eller omgivelser, som kan skjule vigtige mønstre. En denoising autoencoder er en type neuralt netværk, der lærer at genskabe en ren version af et signal ud fra en støjfyldt udgave. Afhandlingen undersøger anvendelsen af denoising autoencoders til at rense biosignaler og beskriver centrale overvejelser som at bevare relevante signaldetaljer og undgå at fjerne klinisk vigtig information. Formålet er at gøre biosignaler tydeligere og mere brugbare i efterfølgende analyser.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords