Abstract

 The ultimate goal of Galactic Archaeology is to use stars as tracers to unravel the evolutionary history of our own galaxy, The Milky Way. Galactic Archaeology today is a fundamentally different field than it was just a decade ago. The order of magnitude of change has been powered by our ability to gather extraordinarily large data sets from ever more powerful instruments. With big data, Galactic archaeology has also taken a quantum leap forward in its relationship with statistics and machine learning. In this talk, I will discuss how the characterization of the chemical and kinematic properties of stars can unveil the subtle engines that drive the evolution of the Milky Way and whose exhaust leaves traces in gentle variations that are now perceptible with statistics. I will also demonstrate how machine learning could alleviate many impasses that bottleneck Galactic Archaeology. In particular, I will illustrate how to model Galactic dynamics with a non-parametric gravitational potential. I will also discuss a better-characterization of non-Gaussianity and auto-calibration of imperfect synthetic simulations with large datasets via machine learning.