MACHINE LEARNING TECHNIQUES APPLICATION TO ASTROMETRY IN THE GAIA MISSION - NELLY GAILLARD

The data processing task of the Gaia mission is large and complex. One of its central elements is the Astrometric Global Iterative Solution (AGIS), which produces and delivers the core astrometry data products.
One of the most challenging tasks that we need to tackle in the software producing Gaia’s astrometric solution is creating a calibration model accurate enough to consider the subtle effects, since they may have an impact on the quality of the solution at the micro-arcsecond level.
Among AGIS related data, the first to be analysed are the post-fit residuals. These are the differences between the observations and the predictions obtained using the AGIS source, attitude and calibration model. Up to now, they have been manually explored, by plotting them in different partitions to identify anomalous clusters of points. This task is time consuming and lends itself to automated analysis by means of machine learning techniques, which could make the procedure more efficient and systematic.
With the objective of performing anomaly detection on the residuals, the first task is to define what an anomaly is, which has been done exploring different data visualization and feature engineering techniques. In this poster we will present an overview of the project and its status, highlighting possible ways forward.