Denoising medium resolution stellar spectra with U-Net convolutional neural networks
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917975759519744 |
|---|---|
| author | Pál, Balázs Dobos, László |
| author_facet | Pál, Balázs Dobos, László |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_02523 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Denoising medium resolution stellar spectra with U-Net convolutional neural networks Pál, Balázs Dobos, László Instrumentation and Methods for Astrophysics Solar and Stellar Astrophysics 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. |
| title | Denoising medium resolution stellar spectra with U-Net convolutional neural networks |
| topic | Instrumentation and Methods for Astrophysics Solar and Stellar Astrophysics |
| url | https://arxiv.org/abs/2504.02523 |