Context. The availability of large bandwidth receivers for millimeter radio
telescopes allows the acquisition of position-position-frequency data cubes
over a wide field of view and a broad frequency coverage. These cubes contain
much information on the physical, chemical, and kinematical properties of the
emitting gas. However, their large size coupled with inhomogenous
signal-to-noise ratio (SNR) are major challenges for consistent analysis and
interpretation.Aims. We search for a denoising method of the low SNR regions of
the studied data cubes that would allow to recover the low SNR emission without
distorting the signals with high SNR.Methods. We perform an in-depth data
analysis of the 13 CO and C 17 O (1 -- 0) data cubes obtained as part of the
ORION-B large program performed at the IRAM 30m telescope. We analyse the
statistical properties of the noise and the evolution of the correlation of the
signal in a given frequency channel with that of the adjacent channels. This
allows us to propose significant improvements of typical autoassociative neural
networks, often used to denoise hyperspectral Earth remote sensing data.
Applying this method to the 13 CO (1 -- 0) cube, we compare the denoised data
with those derived with the multiple Gaussian fitting algorithm ROHSA,
considered as the state of the art procedure for data line cubes.Results. The
nature of astronomical spectral data cubes is distinct from that of the
hyperspectral data usually studied in the Earth remote sensing literature
because the observed intensities become statistically independent beyond a
short channel separation. This lack of redundancy in data has led us to adapt
the method, notably by taking into account the sparsity of the signal along the
spectral axis. The application of the proposed algorithm leads to an increase
of the SNR in voxels with weak signal, while preserving the spectral shape of
the data in high SNR voxels.Conclusions. The proposed algorithm that combines a
detailed analysis of the noise statistics with an innovative autoencoder
architecture is a promising path to denoise radio-astronomy line data cubes. In
the future, exploring whether a better use of the spatial correlations of the
noise may further improve the denoising performances seems a promising avenue.
In addition,
Description
Deep learning denoising by dimension reduction: Application to the ORION-B line cubes
%0 Generic
%1 einig2023learning
%A Einig, Lucas
%A Pety, Jérôme
%A Roueff, Antoine
%A Vandame, Paul
%A Chanussot, Jocelyn
%A Gerin, Maryvonne
%A Orkisz, Jan H.
%A Palud, Pierre
%A Santa-Maria, Miriam Garcia
%A Magalhaes, Victor de Souza
%A Bešlić, Ivana
%A Bardeau, Sébastien
%A Bron, Emeric E.
%A Chainais, Pierre
%A Goicoechea, Javier R
%A Gratier, Pierre
%A Veloso, Viviana Guzman
%A Hughes, Annie
%A Kainulainen, Jouni
%A Languignon, David
%A Lallement, Rosine
%A Levrier, François
%A Lis, Dariuscz C.
%A Liszt, Harvey
%A Bourlot, Jacques Le
%A Petit, Franck Le
%A Öberg, Karin Danielsson
%A Peretto, Nicolas
%A Roueff, Evelyne
%A Sievers, Albrecht
%A Thouvenin, Pierre-Antoine
%A Tremblin, Pascal
%D 2023
%K data
%T Deep learning denoising by dimension reduction: Application to the
ORION-B line cubes
%U http://arxiv.org/abs/2307.13009
%X Context. The availability of large bandwidth receivers for millimeter radio
telescopes allows the acquisition of position-position-frequency data cubes
over a wide field of view and a broad frequency coverage. These cubes contain
much information on the physical, chemical, and kinematical properties of the
emitting gas. However, their large size coupled with inhomogenous
signal-to-noise ratio (SNR) are major challenges for consistent analysis and
interpretation.Aims. We search for a denoising method of the low SNR regions of
the studied data cubes that would allow to recover the low SNR emission without
distorting the signals with high SNR.Methods. We perform an in-depth data
analysis of the 13 CO and C 17 O (1 -- 0) data cubes obtained as part of the
ORION-B large program performed at the IRAM 30m telescope. We analyse the
statistical properties of the noise and the evolution of the correlation of the
signal in a given frequency channel with that of the adjacent channels. This
allows us to propose significant improvements of typical autoassociative neural
networks, often used to denoise hyperspectral Earth remote sensing data.
Applying this method to the 13 CO (1 -- 0) cube, we compare the denoised data
with those derived with the multiple Gaussian fitting algorithm ROHSA,
considered as the state of the art procedure for data line cubes.Results. The
nature of astronomical spectral data cubes is distinct from that of the
hyperspectral data usually studied in the Earth remote sensing literature
because the observed intensities become statistically independent beyond a
short channel separation. This lack of redundancy in data has led us to adapt
the method, notably by taking into account the sparsity of the signal along the
spectral axis. The application of the proposed algorithm leads to an increase
of the SNR in voxels with weak signal, while preserving the spectral shape of
the data in high SNR voxels.Conclusions. The proposed algorithm that combines a
detailed analysis of the noise statistics with an innovative autoencoder
architecture is a promising path to denoise radio-astronomy line data cubes. In
the future, exploring whether a better use of the spatial correlations of the
noise may further improve the denoising performances seems a promising avenue.
In addition,
@misc{einig2023learning,
abstract = {Context. The availability of large bandwidth receivers for millimeter radio
telescopes allows the acquisition of position-position-frequency data cubes
over a wide field of view and a broad frequency coverage. These cubes contain
much information on the physical, chemical, and kinematical properties of the
emitting gas. However, their large size coupled with inhomogenous
signal-to-noise ratio (SNR) are major challenges for consistent analysis and
interpretation.Aims. We search for a denoising method of the low SNR regions of
the studied data cubes that would allow to recover the low SNR emission without
distorting the signals with high SNR.Methods. We perform an in-depth data
analysis of the 13 CO and C 17 O (1 -- 0) data cubes obtained as part of the
ORION-B large program performed at the IRAM 30m telescope. We analyse the
statistical properties of the noise and the evolution of the correlation of the
signal in a given frequency channel with that of the adjacent channels. This
allows us to propose significant improvements of typical autoassociative neural
networks, often used to denoise hyperspectral Earth remote sensing data.
Applying this method to the 13 CO (1 -- 0) cube, we compare the denoised data
with those derived with the multiple Gaussian fitting algorithm ROHSA,
considered as the state of the art procedure for data line cubes.Results. The
nature of astronomical spectral data cubes is distinct from that of the
hyperspectral data usually studied in the Earth remote sensing literature
because the observed intensities become statistically independent beyond a
short channel separation. This lack of redundancy in data has led us to adapt
the method, notably by taking into account the sparsity of the signal along the
spectral axis. The application of the proposed algorithm leads to an increase
of the SNR in voxels with weak signal, while preserving the spectral shape of
the data in high SNR voxels.Conclusions. The proposed algorithm that combines a
detailed analysis of the noise statistics with an innovative autoencoder
architecture is a promising path to denoise radio-astronomy line data cubes. In
the future, exploring whether a better use of the spatial correlations of the
noise may further improve the denoising performances seems a promising avenue.
In addition,},
added-at = {2023-08-03T10:10:12.000+0200},
author = {Einig, Lucas and Pety, Jérôme and Roueff, Antoine and Vandame, Paul and Chanussot, Jocelyn and Gerin, Maryvonne and Orkisz, Jan H. and Palud, Pierre and Santa-Maria, Miriam Garcia and Magalhaes, Victor de Souza and Bešlić, Ivana and Bardeau, Sébastien and Bron, Emeric E. and Chainais, Pierre and Goicoechea, Javier R and Gratier, Pierre and Veloso, Viviana Guzman and Hughes, Annie and Kainulainen, Jouni and Languignon, David and Lallement, Rosine and Levrier, François and Lis, Dariuscz C. and Liszt, Harvey and Bourlot, Jacques Le and Petit, Franck Le and Öberg, Karin Danielsson and Peretto, Nicolas and Roueff, Evelyne and Sievers, Albrecht and Thouvenin, Pierre-Antoine and Tremblin, Pascal},
biburl = {https://www.bibsonomy.org/bibtex/29313cd59f8d855b72feccb472fb06b0f/fboulang},
description = {Deep learning denoising by dimension reduction: Application to the ORION-B line cubes},
interhash = {5738f138482029165179aa76255e7a06},
intrahash = {9313cd59f8d855b72feccb472fb06b0f},
keywords = {data},
note = {cite arxiv:2307.13009},
timestamp = {2023-08-03T10:10:12.000+0200},
title = {Deep learning denoising by dimension reduction: Application to the
ORION-B line cubes},
url = {http://arxiv.org/abs/2307.13009},
year = 2023
}