|
EPoS |
|
EPoS Contribution
|
|
Searching for young stellar objects through SEDs by machine learning
Shih-Ping Lai NTHU, Hsinchu, TW | |
| Accurate measurements of statistical properties, such as the star formation rate and the lifetime of young stellar objects (YSOs) in different stages, are essential for constraining star formation theories. However, distinguishing between galaxies and YSOs based solely on spectral energy distributions (SEDs) is a challenging task. Here, we present the Spectrum Classifier of Astronomical Objects (SCAO), based on a Fully Connected Neural Network (FCN), designed for classifying regular stars, galaxies, and YSOs. In contrast to previous classifiers, SCAO is trained exclusively on high-quality data from the Molecular Cores to Planetforming Disks (c2d) catalog with reliable labels, without relying on a priori theoretical knowledge. SCAO demonstrates exceptional performance, achieving high precision (>96%) and recall (>98%) rates for YSOs with just eight included bands. Applying SCAO to Spitzer Enhanced Imaging Products (SEIP), the most comprehensive Spitzer observations catalog, identifies 129,219 YSO candidates. We further transfered our YSO models to the ALLWISE catalog and identified ~1.2 million YSO candidates across the entired sky. | |
 | |
| Caption: Can machine learning help us distinguish YSOs and background galaxies? (SCAO is accessible at http://scao.astr.nthu.edu.tw) | |
| Collaborators: Y.-L. Chiu, NTHU, TW Y.-C. Hu, NTHU, TW |
Key publication
Relevant topic(s): Core MF Molecular Clouds |
| Relevant Big Question: How can machine learning contribute to the identification of Young Stellar Objects and advance our understanding of the underlying physical processes involved? | |