B1722
Title: Sparsity-based statistical methods for the estimation of the cosmological microwave background
Authors: Jerome Bobin - CEA Saclay (France) [presenting]
Abstract: The Cosmological Microwave Background (CMB) is, by far, the oldest observable signal in the universe that has been emitted more than 13 billions years ago. This relic signal is of paramount importancefor cosmologists since it carries invaluable information about the dawn of our Universe and its evolution. However, accessing such signal is a highly challenging estimationproblem that requires the development of dedicated statistical methods. Indeed, the CMB can only be observed from multispectral data, through complex mixtures of various kinds of contaminants that include astrophysical components,noise, instrumental effect, etc. We first show how extracting one of the most important signal in cosmology can be recast as a challenging inverse problem in statistics: blind source separation. In this context, we emphasize on how modern-day sparse signal models and methods allowed for the very accurate estimation of the CMB from the latest data delivered by the Planck space mission of the European Space Agency.
Journal