Résumé / Abstract Journal-club_Univers

Journal-club Univers / Journal-club Universe

« From detecting strong gravitational lenses to modeling intrinsic alignments with Deep Learning »

François Lanusse
Univ. Carnegie Mellon (Pittsburgh, Pennsylvanie, Etats-Unis d'Amérique)

The upcoming generation of cosmological surveys such as LSST or Euclid will aim at shedding some much needed light on the physical nature of dark energy and dark matter by mapping the Universe in great detail and on an unprecedented scale. This involves however new and outstanding challenges at every step of the science analysis, from image processing to the modeling of astrophysical systematics. In this talk I will illustrate how recent advances in Deep Learning open new perspectives for addressing some of theses challenges. In particular, I will present our work on automated strong gravitational lens detection, a problem where deep learning may essentially eliminate the need for human visual inspection (which would have intractable at the scale of LSST). And as a second example of applications, I will illustrate how data driven deep generative models can be used to complement a physical modeling for producing large volume mock galaxy catalogs with realistic intrinsic alignments, with properties learned from hydrodynamical simulations.
mardi 6 mars 2018 - 11:15
Salle 281, Institut d'Astrophysique
Pages web du journal-club / Journal-club's webpage