Search Menu
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information
Search Scholarly Publications

Title: DeepCMB: Lensing reconstruction of the cosmic microwave background with deep neural networks

Journal Article · · Astronomy and Computing
ORCiD logo [1];  [2];  [3];  [4];  [5];  [6]
  1. Univ. of Chicago, IL (United States). Enrico Fermi Inst. and Kadanoff Center for Theoretical Physics; Fermi National Accelerator Lab. (FNAL), Batavia, IL (United States)
  2. Univ. of Chicago, IL (United States). Kavli Inst. for Cosmological Physics (KICP)
  3. Univ. of Chicago, IL (United States). Kavli Inst. for Cosmological Physics (KICP); Univ. of Chicago, IL (United States). Dept. of Astronomy and Astropysics; Fermi National Accelerator Lab. (FNAL), Batavia, IL (United States)
  4. Univ. of Chicago, IL (United States). Enrico Fermi Inst. and Kadanoff Center for Theoretical Physics; Univ. of Chicago, IL (United States). Kavli Inst. for Cosmological Physics (KICP)
  5. Brown Univ., Providence, RI (United States). Inst. for Computational and Experimental Research in Mathematics
  6. Descartes Labs, Santa Fe, CA (United States)
+ Show Author Affiliations

Next-generation cosmic microwave background (CMB) experiments will have lower noise and therefore increased sensitivity, enabling improved constraints on fundamental physics parameters such as the sum of neutrino masses and the tensor-to-scalar ratio r. Achieving competitive constraints on these parameters requires high signal-to-noise extraction of the projected gravitational potential from the CMB maps. Standard methods for reconstructing the lensing potential employ the quadratic estimator (QE). However, the QE performs suboptimally at the low noise levels expected in upcoming experiments. Other methods, like maximum likelihood estimators (MLE), are under active development. In this work, we demonstrate reconstruction of the CMB lensing potential with deep convolutional neural networks (CNN) - ie, a ResUNet. The network is trained and tested on simulated data, and otherwise has no physical parametrization related to the physical processes of the CMB and gravitational lensing. We show that, over a wide range of angular scales, ResUNets recover the input gravitational potential with a higher signal-to-noise ratio than the QE method, reaching levels comparable to analytic approximations of MLE methods. We demonstrate that the network outputs quantifiably different lensing maps when given input CMB maps generated with different cosmologies. We also show we can use the reconstructed lensing map for cosmological parameter estimation. This application of CNN provides a few innovations at the intersection of cosmology and machine learning. First, while training and regressing on images, we predict a continuous-variable field rather than discrete classes. Second, we are able to establish uncertainty measures for the network output that are analogous to standard methods. We expect this approach to excel in capturing hard-to-model non-Gaussian astrophysical foreground and noise contributions.

Research Organization:
Fermi National Accelerator Laboratory (FNAL), Batavia, IL (United States)
Sponsoring Organization:
USDOE Office of Science (SC), High Energy Physics (HEP); USDOE
Grant/Contract Number:
AC02-07CH11359
OSTI ID:
1487051
Alternate ID(s):
OSTI ID: 1780551
Report Number(s):
arXiv:1810.01483; FERMILAB-PUB-18-515-A-CD; oai:inspirehep.net:1696852
Journal Information:
Astronomy and Computing, Vol. 28, Issue C; ISSN 2213-1337
Publisher:
ElsevierCopyright Statement
Country of Publication:
United States
Language:
English

References (43)

Measurement of the Cosmic Microwave Background Polarization Lensing Power Spectrum with the POLARBEAR Experiment journal July 2014
NINE-YEAR WILKINSON MICROWAVE ANISOTROPY PROBE ( WMAP ) OBSERVATIONS: FINAL MAPS AND RESULTS journal September 2013
Constraints on Primordial Gravitational Waves Using P l a n c k , WMAP, and New BICEP2/ K e c k Observations through the 2015 Season journal November 2018
Bicep2/ KECK ARRAY VIII: MEASUREMENT OF GRAVITATIONAL LENSING FROM LARGE-SCALE B -MODE POLARIZATION journal December 2016
Improved Constraints on Cosmology and Foregrounds from BICEP2 and Keck Array Cosmic Microwave Background Data with Inclusion of 95 GHz Band journal January 2016
Learning the hidden structure of speech journal April 1988
Deep Residual Learning for Image Recognition conference June 2016
Measurements of the Temperature and E-mode Polarization of the CMB from 500 Square Degrees of SPTpol Data journal January 2018
NINE-YEAR WILKINSON MICROWAVE ANISOTROPY PROBE ( WMAP ) OBSERVATIONS: COSMOLOGICAL PARAMETER RESULTS journal September 2013
Reducing the Dimensionality of Data with Neural Networks journal July 2006
Reconstruction of lensing from the cosmic microwave background polarization journal October 2003
Weak lensing of the CMB: A harmonic approach journal July 2000
Mass Reconstruction with Cosmic Microwave Background Polarization journal August 2002
A Probe of Primordial Gravity Waves and Vorticity journal March 1997
MEASUREMENTS OF SUB-DEGREE B -MODE POLARIZATION IN THE COSMIC MICROWAVE BACKGROUND FROM 100 SQUARE DEGREES OF SPTPOL DATA journal July 2015
Lensing reconstruction with CMB temperature and polarization journal June 2003
Cosmological parameters from the first results of Boomerang journal January 2001
Weak gravitational lensing of the CMB journal June 2006
Efficient Computation of Cosmic Microwave Background Anisotropies in Closed Friedmann‐Robertson‐Walker Models journal August 2000
The Atacama Cosmology Telescope: two-season ACTPol spectra and parameters journal June 2017
CMB Polarization B -mode Delensing with SPTpol and Herschel journal August 2017
Measurement of the cosmic microwave background spectrum by the COBE FIRAS instrument journal January 1994
Mission Design of LiteBIRD journal January 2014
A high-bias, low-variance introduction to Machine Learning for physicists journal May 2019
Bayesian delensing of CMB temperature and polarization journal July 2019
A 2500 deg 2 CMB Lensing Map from Combined South Pole Telescope and Planck Data journal November 2017
DeepSphere: Efficient spherical convolutional neural network with HEALPix sampling for cosmological applications journal April 2019
Planck 2015 results : XIII. Cosmological parameters journal September 2016
Planck 2015 results : XV. Gravitational lensing journal September 2016
A Measurement of the Cosmic Microwave Background B -mode Polarization Power Spectrum at Subdegree Scales from Two Years of polarbear Data journal October 2017
U-Net: Convolutional Networks for Biomedical Image Segmentation
  • Ronneberger, Olaf; Fischer, Philipp; Brox, Thomas
  • Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III https://doi.org/10.1007/978-3-319-24574-4_28
book November 2015
Signature of Gravity Waves in the Polarization of the Microwave Background journal March 1997
Two-season Atacama Cosmology Telescope polarimeter lensing power spectrum journal June 2017
Mastering the game of Go without human knowledge journal October 2017
Delensing CMB polarization with external datasets journal June 2012
A Measurement of the Cosmic Microwave Background Gravitational Lensing Potential from 100 Square Degrees of Sptpol data journal August 2015
The Simons Observatory: science goals and forecasts journal February 2019
A Guide to Designing Future Ground-Based Cosmic Microwave Background Experiments journal June 2014
A guide to convolution arithmetic for deep learning preprint January 2016
Self-Normalizing Neural Networks text January 2017
Bayesian delensing of CMB temperature and polarization text January 2017
Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge preprint January 2017
Averaging Weights Leads to Wider Optima and Better Generalization preprint January 2018

Cited By (2)

Learning to predict the cosmological structure formation journal June 2019
Learning to Predict the Cosmological Structure Formation text January 2018

Figures / Tables (12)

Figure 1(p. 5)figure Figure 1
Figure 2(p. 7)figure Figure 2
Figure 3(p. 7)figure Figure 3
Figure 4(p. 8)figure Figure 4
Figure 5(p. 9)figure Figure 5
Figure 6(p. 9)figure Figure 6
Figure 7(p. 10)figure Figure 7
Figure 8(p. 11)figure Figure 8
Figure 10(p. 12)figure Figure 10
Figure 9(p. 12)figure Figure 9
Figure 11(p. 13)figure Figure 11
Figure 12(p. 15)figure Figure 12

Similar Records

Related Subjects

79 ASTRONOMY AND ASTROPHYSICS
cosmic microwave background
cosmology
deep learning
convolutional neural networks