Ponente
Descripción
The study of the cosmic microwave background (CMB) and its anisotropies is crucial for understanding the cosmology and formation processes of the Universe. Investigating the E and B modes of CMB polarization provides insights into early Universe density and gravitational perturbations, respectively. B-mode polarization, in particular, offers vital information regarding the existence of primordial gravitational waves and serves as a test for inflationary cosmological models, though it poses significant challenges for characterization.
In this study, we simulate the CMB across various frequencies and tensor perturbation intensities, accounting for contamination from galactic foreground signals (thermal and synchrotron emission), extragalactic point sources, and instrumental noise. We develop a Fully Convolutional Neural Network (FCNN) capable of recovering the CMB with the aim of determining up to which value of the tensor-to-scalar ratio this method is able to retrive the signal. This tool aims to enhance next-generation high-resolution experiments for the precise extraction of the CMB signal and the characterization of background components.
This presentation will provide an overview of the problem, outline our objectives, and detail the development and results of the implemented artificial intelligence approach.
