Year: 2025
Type: preprint
Abstract: We investigated the use of a U-Net convolutional neural network for denoising simulated medium-resolution spectroscopic observations of stars. Simulated spectra were generated under realistic observational conditions resembling the Subaru Prime Focus Spectrograph (PFS). We found that our U-Net model effectively captured spectral features, achieving an average relative error of around $1\%$ across a broad range of stellar parameters, despite a limited training set of only $1000$ observations and a relatively short training period. Although U-Net did not reach the performance previously demonstrated by fully-connected denoising autoencoders (DAEs) consisting of dense layers and trained extensively on larger datasets, it outperformed dense networks trained under similarly constrained conditions. These results indicate that the U-Net architecture offers rapid, robust feature learning and may be particularly advantageous in scenarios involving initial denoising, subsequently refined by more accurate, but otherwise slower deep-learning models.
Source: arXiv (Cornell University)
Authors: Pál, Balázs, Dobos, László
Institution: Eötvös Loránd University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funders: Johns Hopkins University, European Commission, Hungarian Scientific Research Fund, Nemzeti Kutatási Fejlesztési és Innovációs Hivatal, Nemzeti Kutatási, Fejlesztési és Innovaciós Alap, Wigner Fizikai Kutatóközpont, Magyar Tudományos Akadémia, National Research, Development and Innovation Office
OpenAlex ID: W4410289643