An Unsupervised Learning Method for Radio Interferometry Deconvolution
Given the incomplete sampling of spatial frequencies by radio interferometers, achieving precise restoration of astrophysical information remains challenging. To address this ill-posed problem, compressive sensing (CS) provides a robust framework for stable and unique recovery of sky brightness dist...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2025-01-01
|
| Series: | The Astrophysical Journal Supplement Series |
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-4365/add1b7 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849722753248133120 |
|---|---|
| author | Lei Yu Bin Liu Cheng-Jin Jin Ru-Rong Chen Hong-Wei Xi Bo Peng |
| author_facet | Lei Yu Bin Liu Cheng-Jin Jin Ru-Rong Chen Hong-Wei Xi Bo Peng |
| author_sort | Lei Yu |
| collection | DOAJ |
| description | Given the incomplete sampling of spatial frequencies by radio interferometers, achieving precise restoration of astrophysical information remains challenging. To address this ill-posed problem, compressive sensing (CS) provides a robust framework for stable and unique recovery of sky brightness distributions in noisy environments, contingent upon satisfying specific conditions. We explore the applicability of CS theory and find that for radio interferometric telescopes, the conditions can be simplified to sparse representation. Building on this insight, we develop a deep dictionary (realized through a convolutional neural network), which is designed to be multiresolution and overcomplete, to achieve sparse representation and integrate it within the CS framework. The resulting method is a novel, fully interpretable unsupervised learning approach that combines the mathematical rigor of CS with the expressive power of deep neural networks, effectively bridging the gap between deep learning and classical dictionary methods. During the deconvolution process, the model image and the deep dictionary are updated alternatively. This approach enables efficient and accurate recovery of extended sources with complex morphologies from noisy measurements. Comparative analyses with state-of-the-art algorithms demonstrate the outstanding performance of our method, i.e., achieving a dynamic range nearly 45–100 times higher than that of multiscale CLEAN. |
| format | Article |
| id | doaj-art-2582a13640974dda89fbe84670e22845 |
| institution | DOAJ |
| issn | 0067-0049 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | The Astrophysical Journal Supplement Series |
| spelling | doaj-art-2582a13640974dda89fbe84670e228452025-08-20T03:11:14ZengIOP PublishingThe Astrophysical Journal Supplement Series0067-00492025-01-0127825410.3847/1538-4365/add1b7An Unsupervised Learning Method for Radio Interferometry DeconvolutionLei Yu0https://orcid.org/0009-0005-2954-0910Bin Liu1https://orcid.org/0000-0002-1311-8839Cheng-Jin Jin2Ru-Rong Chen3Hong-Wei Xi4https://orcid.org/0000-0001-6642-8307Bo Peng5https://orcid.org/0000-0001-6956-6553CAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; University of Chinese Academy of Sciences , Beijing, 100049, People’s Republic of China; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of ChinaCAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of ChinaCAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of ChinaCAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of ChinaCAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of ChinaCAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences , Beijing 100101, People’s Republic of China ; yulei@nao.cas.cn, bliu@nao.cas.cn, pb@nao.cas.cn; State Key Laboratory of Radio Astronomy and Technology , Beijing 100101, People's Republic of China; Department of Astronomy and Institute of Interdisciplinary Studies, Hunan Normal University , Changsha, Hunan 410081, People’s Republic of ChinaGiven the incomplete sampling of spatial frequencies by radio interferometers, achieving precise restoration of astrophysical information remains challenging. To address this ill-posed problem, compressive sensing (CS) provides a robust framework for stable and unique recovery of sky brightness distributions in noisy environments, contingent upon satisfying specific conditions. We explore the applicability of CS theory and find that for radio interferometric telescopes, the conditions can be simplified to sparse representation. Building on this insight, we develop a deep dictionary (realized through a convolutional neural network), which is designed to be multiresolution and overcomplete, to achieve sparse representation and integrate it within the CS framework. The resulting method is a novel, fully interpretable unsupervised learning approach that combines the mathematical rigor of CS with the expressive power of deep neural networks, effectively bridging the gap between deep learning and classical dictionary methods. During the deconvolution process, the model image and the deep dictionary are updated alternatively. This approach enables efficient and accurate recovery of extended sources with complex morphologies from noisy measurements. Comparative analyses with state-of-the-art algorithms demonstrate the outstanding performance of our method, i.e., achieving a dynamic range nearly 45–100 times higher than that of multiscale CLEAN.https://doi.org/10.3847/1538-4365/add1b7Radio interferometryDeconvolutionNeural networksAstronomy image processing |
| spellingShingle | Lei Yu Bin Liu Cheng-Jin Jin Ru-Rong Chen Hong-Wei Xi Bo Peng An Unsupervised Learning Method for Radio Interferometry Deconvolution The Astrophysical Journal Supplement Series Radio interferometry Deconvolution Neural networks Astronomy image processing |
| title | An Unsupervised Learning Method for Radio Interferometry Deconvolution |
| title_full | An Unsupervised Learning Method for Radio Interferometry Deconvolution |
| title_fullStr | An Unsupervised Learning Method for Radio Interferometry Deconvolution |
| title_full_unstemmed | An Unsupervised Learning Method for Radio Interferometry Deconvolution |
| title_short | An Unsupervised Learning Method for Radio Interferometry Deconvolution |
| title_sort | unsupervised learning method for radio interferometry deconvolution |
| topic | Radio interferometry Deconvolution Neural networks Astronomy image processing |
| url | https://doi.org/10.3847/1538-4365/add1b7 |
| work_keys_str_mv | AT leiyu anunsupervisedlearningmethodforradiointerferometrydeconvolution AT binliu anunsupervisedlearningmethodforradiointerferometrydeconvolution AT chengjinjin anunsupervisedlearningmethodforradiointerferometrydeconvolution AT rurongchen anunsupervisedlearningmethodforradiointerferometrydeconvolution AT hongweixi anunsupervisedlearningmethodforradiointerferometrydeconvolution AT bopeng anunsupervisedlearningmethodforradiointerferometrydeconvolution AT leiyu unsupervisedlearningmethodforradiointerferometrydeconvolution AT binliu unsupervisedlearningmethodforradiointerferometrydeconvolution AT chengjinjin unsupervisedlearningmethodforradiointerferometrydeconvolution AT rurongchen unsupervisedlearningmethodforradiointerferometrydeconvolution AT hongweixi unsupervisedlearningmethodforradiointerferometrydeconvolution AT bopeng unsupervisedlearningmethodforradiointerferometrydeconvolution |