Machine learning search for Gaia DR3 astrometric exoplanet orbits - SSW

MACHINE LEARNING SEARCH FOR GAIA DR3 ASTROMETRIC EXOPLANET ORBITS - PABLO GÓMEZ

The third Gaia data release (GDR3) contains ~170,000 astrometric orbit solutions of two-body systems located within ~500 pc of the Sun. The determination of the component masses of these systems usually hinges on incorporating complementary observations in addition to the astrometry, e.g., spectroscopy and radial velocities. Several GDR3 two-body systems with exoplanet, brown-dwarf, stellar, and black-hole components have been confirmed in this way. Using ESA Datalabs, we developed an alternative machine learning approach that uses only the GDR3 orbital solutions with the aim of identifying the best candidates for exoplanets and brown-dwarf companions. Based on confirmed substellar companions in the literature, we use semi-supervised anomaly detection methods in combination with extreme gradient boosting and random forest classifiers to determine likely low-mass outliers in the population of non-single sources. We employ and study feature importance to investigate the method's plausibility and produced a list of likely candidates for further study. Our preliminary findings suggest that this new approach is a powerful complement to the traditional and “manual” identification of substellar-companion candidates in Gaia astrometric orbits. It is particularly relevant in the context of GDR4 and its expected exoplanet discovery yield.