Machine Learning and Gaia DR2, on the hunt for open clusters

Alfred Castro Ginard
Universidad de Barcelona

Abstract:

The second release of Gaia data (Gaia DR2), containing precise astrometry and excellent photometry for more than 1.3 billion sources, opened many new possibilities for Big Data in Astronomy. Machine Learning techniques have proved to be an indispensable tool for the analysis of such a wealth of data. The field of open clusters is the perfect testbed for these techniques. From unsupervised learning algorithms, able to find unknown structures in a N-dimensional data set, i.e. detect stars with very similar position, parallax and proper motions; to supervised learning algorithms which can learn how to recognise a given pattern in the data, for instance the isochrone pattern cluster member stars follow in a color-magnitude diagram. In this talk, I will go through how a combination of these Machine Learning techniques can boost the detection of open clusters in Gaia DR2, and which are the steps to adapt existing methodologies to a more general search.