Exploring the unassociated Galactic 4FGL-DR3 sources using multiwavelength machine learning classification (MUWCLASS)

The unassociated sources comprise about half of the Galactic (with |b| < 10°) gamma-ray sources in the 4FGL-DR3 catalog. The gamma-ray properties alone are often insufficient to establish the nature of the source while the large 4FGL positional uncertainties make it challenging to find the counterparts at lower energies. We examined 114 4FGL-DR3 unassociated Galactic sources (hereafter 4FGLUGSs) which are covered by Chandra Source Catalog v2.0 (CSCv2.0) and extracted the properties of all X-ray sources found within the error ellipses of 4FGLUGSs. Using the Multiwavelength Machine Learning Classification Pipeline (MUWCLASS) developed by our group (see other talk ID 39) we obtained classifications of the X-ray sources by combining optical, infrared and X-ray information for each source. MUWCLASS uses a training dataset of about 3,000 classified X-ray sources and the supervised learning random forest algorithm to probabilistically determine classes for the X-ray sources in question. I will present the results of the classifications of potential X-ray counterparts of 4FGLUGSs which can be used to identify the nature of the gamma-ray sources. I will also demonstrate a publicly available online tool which allows for easy interactive visualization and exploration of the 4FGL-DR3 catalog and classification results.