I will present my work on the population study of the Milky Way’s X-ray emitting stellar objects by combining the latest XMM-Newton (4XMM-DR13) observations with Gaia (DR3) astrometry and photometry. Using the probabilistic ARCHES cross-matching tool developed at the Observatory of Strasbourg, we identified 37,901 high-confidence optical stellar counterparts to XMM sources. The CMD revealed a prominent second sequence, which Simbad classifications showed to be dominated by magnetically active young stars and unresolved binaries.
To extend classifications beyond the limited Simbad labels, we trained a deep neural network to distinguish Binaries, Evolved, Main Sequence, Peculiar, and Young stars, achieving 77% accuracy. Together, probabilistic cross-matching, multiwavelength data, and machine learning provide an efficient framework for automatically characterizing Galactic X-ray stellar populations.