Classification of Extragalactic X-Ray Sources Using Machin Learning Methods
Abstract
Currently, only a small fraction of extragalactic X-ray sources have reliable classifications. For example, out of the ~2,000 X-ray sources in M31 and M33, only ~25% have been classified. Typically, the X-ray data alone are not enough to reveal the nature of the X-ray source. Therefore, creating an automated machine learning (ML) tool for classification of extragalactic X-ray sources with multi-wavelength data will enable us to understand X-ray source populations in a plethora of nearby galaxies. Modern ML methods can be used to quickly analyze the vast amount of multi-wavelength data for these unclassified sources providing both the classifications and their confidences. We are using data from the Hubble Space Telescope and Chandra X-ray Observatory, to build and test an automated ML classification pipeline. The pipeline makes use of supervised ML methods and relies on a large training dataset. We present the testing and preliminary results of the ML pipeline and discuss the challenges associated with building an automated ML tool for extragalactic purposes.
- Publication:
-
American Astronomical Society Meeting Abstracts #233
- Pub Date:
- January 2019
- Bibcode:
- 2019AAS...23345703R