Presentation #100.60 in the session AGN.
The fourth Fermi-LAT catalog (4FGL) lists 6659 gamma-ray sources. Of these, 4502 gamma-ray sources have been identified/associated with AGNs (3814), pulsars (137), and other sources (551), while 2157 gamma-ray sources (32% of the total) remain unidentified. It has been reported that various types of machine learning and deep learning methods have classified unidentified gamma-ray sources in the Fermi-LAT catalog. In this work we apply kernel Mahalanobis distance methods to spatial, spectral, and time-variation features of gamma-ray sources from 4FGL. We present the classification performance of AGNs, pulsars, and other sources together with the results of the classification of the unidentified gamma-ray sources. Taking advantage of the superior performance of the kernel Mahalanobis distance method in extracting outliers, we also discuss potentially unanticipated gamma-ray sources as dark matter subhalo candidates.