CENN: A fully convolutional neural network for CMB recovery in realistic microwave sky simulations

Abstract

Component separation is the process with which emission sources in astrophysical maps are generally extracted by taking multi-frequency information into account. It is crucial to develop more reliable methods for component separation for future CMB experiments. We aim to develop a new method based on fully convolutional neural networks called the Cosmic microwave background Extraction Neural Network (CENN) in order to extract the CMB signal in total intensity. The frequencies used are the Planck channels 143, 217 and 353 GHz. We validate the network at all sky, and at three latitude intervals: lat1=0^{\circ}<b<5^{\circ}, lat2=5^{\circ}<b<30^{\circ} and lat3=30^{\circ}<b<90^{\circ}, without using any Galactic or point source masks. For training, we make realistic simulations in the form of patches of area 256 pixels, which contain the CMB, Dust, CIB and PS emissions, Sunyaev-Zel'dovich effect and the instrumental noise. After validate the network, we compare the power spectrum from input and output maps. We analyse the power spectrum from the residuals at each latitude interval and at all sky and we study the performance of our model dealing with high contamination at small scales. We obtain a power spectrum with an error of 13{\pm}113 μK^2 for multipoles up to above 4000. For residuals, we obtain 700{\pm}60 μK^2 for lat1, 80{\pm}30 μK^2 for lat2 and 30{\pm}20 μK^2 for lat3. For all sky, we obtain 20{\pm}10 μK^2. We validate the network in a patch with strong contamination at small scales, obtaining an error of 50{\pm}120 μK^2 and residuals of 40{\pm}10 μK^2. Therefore, fully convolutional neural networks are promising methods to perform component separation in future CMB experiments. Particularly, CENN is reliable against different levels of contamination from Galactic and point source foregrounds at both large and small scales.

Accepted for publication in Astronomy & Astrophysics journal. 11 pages, 6 figures

References in corpus (7)

• The Cosmic Infrared Background Resolved by Spitzer. Contributions of Mid-Infrared Galaxies to the Far-Infrared Background
• A full sky, low foreground, high resolution CMB map from WMAP
• Component separation methods for the Planck mission
• Far-infrared to millimeter astrophysical dust emission I: A model based on physical properties of amorphous solids
• Recovering the CMB Signal with Machine Learning
• Single frequency CMB B-mode inference with realistic foregrounds from a single training image
• Bayesian foreground analysis with CMB data

Cited by in corpus (6)

• Recovering Cosmic Microwave Background Polarization Signals with Machine Learning
• Delensing of Cosmic Microwave Background Polarization with machine learning
• Inference of the optical depth to reionization $\tau$ from $\text{Planck}$ CMB maps with convolutional neural networks
• Deep learning approach for identification of HII regions during reionization in 21-cm observations -- III. image recovery
• Constraining the polarisation flux density and angle of point sources by training a convolutional neural network
• Cluster optical depth and pairwise velocity estimation using machine learning