Data-driven dust inference at mid-to-high Galactic latitudes using probabilistic machine learning

ABSTRACT We present a method for accurately and precisely inferring photometric dust reddening towards stars at mid-to-high Galactic latitudes ($|b| > 20^\circ$), using probabilistic machine learning to model the colour–magnitude distribution of zero-extinction stars in these regions. Photometric dust maps rely on a robust method for inferring stellar reddening. At high Galactic latitudes, where extinction is low, such inferences are particularly susceptible to contamination from modelling errors and prior assumptions, potentially introducing artificial structure into dust maps. In this work, we demonstrate the use of normalizing flows to learn the conditional probability distribution of the photometric colour–magnitude relations of zero-extinction stars, conditioned on Galactic cylindrical coordinates for stars at mid-to-high Galactic latitudes. By using the normalizing flow to model the colour–magnitude diagram, we infer the posterior distribution of dust extinction towards stars along different lines of sight by marginalizing over the colour–magnitude flow. We validate our method using data from Gaia, Pan-STARRS, and Two Micron All Sky Surve, showing that we can recover unbiased posteriors and successfully detect dust along the line of sight in two calibration regions at mid-Galactic latitude that have been extensively studied in the context of polarization surveys.