Identifying new X-ray binary candidates in M31 using random forest classification

摘要: Identifying X-ray binary (XRB) candidates in nearby galaxies requires distinguishing them from possible contaminants including foreground stars and background active galactic nuclei. This work investigates the use of supervised machine learning algorithms to identify high-probability candidates. Using a catalogue 943 Chandra sources Andromeda galaxy, we trained tested several classification using properties 163 with previously known types. Amongst tested, find that random forest classifiers give best performance better (XRB/non-XRB) context compared multiple classes. Evaluating our method by comparing classifications visible-light hard observations as part Panchromatic Hubble Treasury, compatibility at 90% level, although caution number common is rather small. The estimated probability an object agrees well between multiclass approaches highest confidence are class. most discriminating bands for 1.7-2.8, 0.5-1.0, 2.0-4.0, 2.0-7.0 keV photon flux ratios. Of 780 unclassified catalogue, 16 new tabulate their follow-up.