Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userInpainting CMB maps using Partial Convolutional Neural Networks
83
0
0.0
(
0
)
Added by
Gabriele Montefalcone
Publication date
2020
fields
Physics
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
We present a novel application of partial convolutional neural networks (PCNN) that can inpaint masked images of the cosmic microwave background. The network can reconstruct both the maps and the power spectra to a few percent for circular and irregularly shaped masks covering up to ~10% of the image area. By performing a Kolmogorov-Smirnov test we show that the reconstructed maps and power spectra are indistinguishable from the input maps and power spectra at the 99.9% level. Moreover, we show that PCNNs can inpaint maps with regular and irregular masks to the same accuracy. This should be particularly beneficial to inpaint irregular masks for the CMB that come from astrophysical sources such as galactic foregrounds. The proof of concept application shown in this paper shows that PCNNs can be an important tool in data analysis pipelines in cosmology.
rate research
Read More
Deep convolutional neural networks have been a popular tool for image generation and restoration. The performance of these networks is related to the capability of learning realistic features from a large dataset. In this work, we applied the problem of inpainting non-Gaussian signal, in the context of Galactic diffuse emissions at the millimetric and sub-millimetric regimes, specifically Synchrotron and Thermal Dust emission. Both of them are affected by contamination at small angular scales due to extra-galactic radio sources (the former) and to dusty star-forming galaxies (the latter). We consider the performances of a nearest-neighbors inpainting technique and compare it with two novels methodologies relying on generative Neural Networks. We show that the generative network is able to reproduce the statistical properties of the ground truth signal more consistently with high confidence level. The Python Inpainter for Cosmological and AStrophysical SOurces (PICASSO) is a package encoding a suite of inpainting methods described in this work and has been made publicly available.
Active Galactic Nuclei (AGN) and star-forming galaxies are leading candidates for being the luminous sources that reionized our Universe. Next-generation 21cm surveys are promising to break degeneracies between a broad range of reionization models, hence revealing the nature of the source population. While many current efforts are focused on a measurement of the 21cm power spectrum, some surveys will also image the 21cm field during reionization. This provides further information with which to determine the nature of reionizing sources. We create a Convolutional Neural Network (CNN) that is efficiently able to distinguish between 21cm maps that are produced by AGN versus galaxies scenarios with an accuracy of 92-100%, depending on redshift and neutral fraction range. An exception to this is when our Universe is highly ionized, since the source models give near-identical 21cm maps in that case. When adding thermal noise from typical 21cm experiments, the classification accuracy depends strongly on the effectiveness of foreground removal. Our results show that if foregrounds can be removed reasonably well, SKA, HERA and LOFAR should be able to discriminate between source models with greater accuracy at a fixed redshift. Only future SKA 21cm surveys are promising to break the degeneracies in the power spectral analysis.
We present a semi-analytical method to investigate the systematic effects and statistical uncertainties of the calculated angular power spectrum when incomplete spherical maps are used. The computed power spectrum suffers in particular a loss of angular frequency resolution, which can be written as delta_l ~ pi/gamma_max, where gamma_max is the effective maximum extent of the partial spherical maps. We propose a correction algorithm to reduce systematic effects on the estimated C_l, as obtained from the partial map projection on the spherical harmonic Ylm(l,m) basis. We have derived near optimal bands and weighting functions in l-space for power spectrum calculation using small maps, and a correction algorithm for partially masked spherical maps that contain information on the angular correlations on all scales.
Advances in radio spectro-polarimetry offer the possibility to disentangle complex regions where relativistic and thermal plasmas mix in the interstellar and intergalactic media. Recent work has shown that apparently simple Faraday Rotation Measure (RM) spectra can be generated by complex sources. This is true even when the distribution of RMs in the complex source greatly exceeds the errors associated with a single component fit to the peak of the Faraday spectrum. We present a convolutional neural network (CNN) that can differentiate between simple Faraday thin spectra and those that contain multiple or Faraday thick sources. We demonstrate that this CNN, trained for the upcoming Polarisation Sky Survey of the Universes Magnetism (POSSUM) early science observations, can identify two component sources 99% of the time, provided that the sources are separated in Faraday depth by $>$10% of the FWHM of the Faraday Point Spread Function, the polarized flux ratio of the sources is $>$0.1, and that the Signal-to-Noise radio (S/N) of the primary component is $>$5. With this S/N cut-off, the false positive rate (simple sources mis-classified as complex) is $<$0.3%. Work is ongoing to include Faraday thick sources in the training and testing of the CNN.
In this work, we present a new, high performance algorithm for background rejection in imaging atmospheric Cherenkov telescopes. We build on the already popular machine-learning techniques used in gamma-ray astronomy by the application of the latest techniques in machine learning, namely recurrent and convolutional neural networks, to the background rejection problem. Use of these machine-learning techniques addresses some of the key challenges encountered in the currently implemented algorithms and helps to significantly increase the background rejection performance at all energies. We apply these machine learning techniques to the H.E.S.S. telescope array, first testing their performance on simulated data and then applying the analysis to two well known gamma-ray sources. With real observational data we find significantly improved performance over the current standard methods, with a 20-25% reduction in the background rate when applying the recurrent neural network analysis. Importantly, we also find that the convolutional neural network results are strongly dependent on the sky brightness in the source region which has important implications for the future implementation of this method in Cherenkov telescope analysis.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
