A grid-based machine learning tool for inferring exoplanet host-star properties.

Uniform inference of exoplanet-host star parameters (such as mass, radius, and age) taking into account the relevant stellar physics is essential in the robust characterisation of planetary systems. This is attributed to the strong synergy between planets and their host-stars. In this poster, we present a machine learning tool which accurately and precisely infers exoplanet host-star parameters, based on available set of atmospheric constraints. The tool rapidly trains on a set of theoretical model observables, delivering instant estimates of stellar parameters. The results from our machine learning tool are consistent with stellar properties of binary stars, LEGACY sample, and sun-like main-sequence stars in the APOKASC catalogue. In particular, our code yield systematics for the binaries of up-to 2.3% in radius and 5.1% in mass. For the LEGACY sample, the systematics in radius, mass, and age are 1.6%, 3.3%, and 14.8%, respectively. Therefore, the tool can be adopted in uniformly deriving stellar properties of a catalogue of exoplanet-host stars.