What question did this study set out to answer?

The aim is to identify very metal-poor stars effectively from large datasets generated by DESI.

A Catalog of High-confidence Very Metal-poor Stars from DESI Based on a Deep Learning Method

View Full Paper

Ask AI

Bookmark

Authors

A novel deep learning framework identifies very metal-poor stars, highlighting significant findings for Galactic studies.

The aim is to identify very metal-poor stars effectively from large datasets generated by DESI.
Developed a dual-model deep learning framework combining 1D-ResNet classifier and regression model.
Utilized transfer learning with high-quality labels from APOGEE and Large Sky Area Telescope.
Optimized classification for DESI spectra with a specific temperature cut to ensure accuracy.
Achieved a classification accuracy of 97.87% and an rms error of 0.093 dex in metallicity predictions.
Constructed a catalog of 2569 high-confidence VMP candidates, including 1377 new contenders.
Validated the reliability of metallicity estimates through internal consistency checks and external cross-matching.

Yang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

View Full Paper

Ask AI

Bookmark

Authors

A novel deep learning framework identifies very metal-poor stars, highlighting significant findings for Galactic studies.

The aim is to identify very metal-poor stars effectively from large datasets generated by DESI.
Developed a dual-model deep learning framework combining 1D-ResNet classifier and regression model.
Utilized transfer learning with high-quality labels from APOGEE and Large Sky Area Telescope.
Optimized classification for DESI spectra with a specific temperature cut to ensure accuracy.
Achieved a classification accuracy of 97.87% and an rms error of 0.093 dex in metallicity predictions.
Constructed a catalog of 2569 high-confidence VMP candidates, including 1377 new contenders.
Validated the reliability of metallicity estimates through internal consistency checks and external cross-matching.

Yang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark