Title: Sparse component separation in astrophysics, with a particular focus on the case of Planck.

Abstract: During the last decade, blind component separation (a.k.a. blind source separation) has played a central role in astrophysics. This is particularly the case of the Planck data, where it has allowed to estimate a highly accurate map of the cosmological microwave background (CMB). Produced 13 billion years ago at the epoch of recombination, CMB is the most ancient signal that is observable so far in our Universe and carries a wealth of information to understand its birth and evolution. During this presentation, I will review the challenges raised by the estimation of the CMB and how dedicated component separation methods have been developed to tackle the estimation of the CMB. If time permits, I will show how the effort made on Planck has been recently extended to radio and X-ray astronomy.