Performance analysis of extragalactic classifications in Gaia Data Release 4

The Discrete Source Classifier (DSC) provides probabilistic classifications of sources in Data Release 4 (GDR4) based on empirically trained Bayesian classifiers. Using the astrometry, photometry, and low-resolution spectra (XP), the DSC classifies all sources into one of three classes: quasar, galaxy, or star. The DSC comprises three trained neural networks and three different combinations of their probabilities. When evaluated as a function of brightness and sky position on a test set that excludes the Magellanic Clouds, the DSC purity in GDR4 has improved but at the cost of a small loss in completeness. The average performance of the best classifiers at magnitudes brighter than G=20 is at least 88% completeness and 96% purity for the extragalactic classes, namely the quasar and galaxy classes. At fainter magnitudes, the performance is lower due to increased noise. The average performance at magnitudes of 20łeqG<20.5 is a minimum of 55% completeness and 71% purity for the extragalactic classes. At G>20.5,mag, the completeness is considerably reduced, primarily for the models that depend on the XP spectra. Furthermore, we trained additional models on the Gaia Gaia Gaia optical data together with mid-infrared photometry from the CatWISE2020 catalogue. Inclusion of infrared photometry increases the completeness of extragalactic samples at G>20,mag between 9 and 29 percentage points at the cost of reducing the purity between 1 and 9 percentage points. In the GDR4, the best DSC-combined classifier prioritising completeness identifies three million quasars and two million galaxies, but with expected high contamination among fainter sources. In contrast, the combined classifiers prioritising purity identify approximately two million quasars and 1.3 million galaxies with an expected lower level of contamination. At the end of this paper, we provide recommendations for enhancing the purity of the DSC extragalactic selection by applying quality cuts to the Gaia photometry and astrometry.