The Variable Universe with the Gaia mission and AI methods

TL;DR

This paper addresses the challenge of extracting robust variability classifications from Gaia's enormous, multi-epoch all-sky dataset. It documents a spectrum of ML-driven approaches, including supervised classification with 28 time-series features and Random Forests, variational autoencoders for latent representations, and citizen-science GaiaVari for validation. It reports DR1-DR3 growth in catalog sizes to tens of millions of variables across dozens of subtypes, and outlines plans for DR4/DR5, including public release of time series from DR4 and cross-survey training with OGLE, ZTF, and TESS. The work positions Gaia as a comprehensive benchmark for time-domain methods and anticipates synergistic data with LSST Rubin Observatory to advance the mapping of the variable sky.

Abstract

The Gaia mission has observed over 2 billion stars repeatedly across the entire sky over 10 years, revealing the many astronomical objects that vary on human timescales from seconds to years. Its repeated astrometric, photometric, spectrophotometric and spectroscopic measurements create an unprecedented dataset to probe the variable celestial sources down to G ~ 21 mag. To extract meaningful results from these many time series for so many sources, we have used machine learning techniques for crossmatching, variability detection, and variability classification. This approach has now led to the largest catalogue of classified variable sources ever produced over the entire celestial sphere.