Automatic Classification of Rare Sources in DFBS
Abstract
The Digitised First Byurakan Survey (DFBS) provides low-resolution dispersion optical spectra for more than 24 million objects. A two-step (rough filter and fine search) machine learning algorithm based on measuring similarities to predefined templates is applied to identify classes of template-like objects in the dataset. The templates include late type stars (carbon and M stars), quasars and white dwarfs. The method can be applied as a quick-look analysis in other sets of low resolution spectra, e.g. in the GAIA mission.
- Publication:
-
Non-Stable Universe: Energetic Resources, Activity Phenomena, and Evolutionary Processes
- Pub Date:
- July 2017
- Bibcode:
- 2017ASPC..511..164T