We are at a tipping point in Astrophysics where fundamentally new science can be empowered by applying machine learning (ML) techniques to large surveys with billions of galaxies and petabytes of data. This combination is the cornerstone of the Marie  Skłodowska-Curie action BESTTIME (grant agreement No. 896225).

Our main objective is to fill the gap bewteen data-driven analysis in the local universe - mostly leveraging the Sloan Digital Sky Survey - and the study of distant galaxies, which until now has been conducted in an old-fashion way, with small number statistics coming from pencil-beam surveys. We have developed new ML tools to measure physical properties of photometric galaxies, and applied these tools to understand their evolution in the early universe, especially stellar mass assembly, star formation quenching, and the relationship between the baryonic components of galaxies and the dark matter halos hosting them.

Such a perspective is perfectly aligned to  oncoming NASA and ESA missions: the  Euclid and Roman space telescopes, designed to observe billions of galaxies, will benefit from the progress made within our project, which has also been instrumental to the realization of the Cosmic Dawn Survey.

BEST TIME results: check out the papers!

Euclid preparation. XI. Mean redshift determination from galaxy redshift probabilities for cosmic shear tomography

Hubble Frontier Field Clusters and Their Parallel Fields: Photometric and Photometric Redshift Catalogs

The Evolving Interstellar Medium of Star-forming Galaxies, as Traced by Stardust

Euclid preparation. XVII. Cosmic Dawn Survey: Spitzer Space Telescope observations of the Euclid deep fields and calibration fields

COSMOS2020: Manifold Learning to Estimate Physical Parameters in Large Galaxy Surveys

COSMOS2020: UV selected galaxies at z≥7.5

COSMOS2020: Cosmic evolution of the stellar-to-halo mass relation for central and satellite galaxies up to z ∼ 5