Cover Picture: Astron. Nachr. 9/2024

Denoising a medium resolution stellar spectrum with neural networks. Top : Example of a noiseless simulated stellar spectrum (blue) transformed to an “observation” by adding a realistic noise component (gray) such that the effective S/N≈19. Middle : Comparison of the original noiseless simulated spe...

Full description

Saved in:
Bibliographic Details
Published in:Astronomische Nachrichten 2024-11, Vol.345 (9-10)
Main Authors: Pál, Balázs, Dobos, László
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Denoising a medium resolution stellar spectrum with neural networks. Top : Example of a noiseless simulated stellar spectrum (blue) transformed to an “observation” by adding a realistic noise component (gray) such that the effective S/N≈19. Middle : Comparison of the original noiseless simulated spectrum (blue) and the reconstructed spectrum (dashed orange) using a trained denoising autoencoder. The two lines overlap almost entirely, indicating the high accuracy of the machine learning method. Bottom : Relative error calculated as the fraction of pixel‐wise residual noise and the original noiseless flux. The mean and maximum of the relative error are 0.175% and 1.806%, respectively. Formore details see the related paper by Pál and Dobos, published in this issue e240049 . image
ISSN:0004-6337
1521-3994
DOI:10.1002/asna.20249017