Compressed Sensing via Measurement-Conditional Generative Models

Pre-trained generators have been frequently adopted in compressed sensing (CS) owing to their ability to effectively estimate signals with the prior of NNs. To further refine the NN-based prior, we propose a framework that allows the generator to utilize additional information from given measurement...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Kyung-Su Kim, Jung Hyun Lee, Eunho Yang
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Compressed sensing artificial neural networks image reconstruction image enhancement signal reconstruction and prediction measurement-conditional generative models Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/bba1dc7b48244d7eade5d2db3dcb8225
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:bba1dc7b48244d7eade5d2db3dcb8225
record_format	dspace
spelling	oai:doaj.org-article:bba1dc7b48244d7eade5d2db3dcb82252021-11-26T00:01:31ZCompressed Sensing via Measurement-Conditional Generative Models2169-353610.1109/ACCESS.2021.3128721https://doaj.org/article/bba1dc7b48244d7eade5d2db3dcb82252021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9617735/https://doaj.org/toc/2169-3536Pre-trained generators have been frequently adopted in compressed sensing (CS) owing to their ability to effectively estimate signals with the prior of NNs. To further refine the NN-based prior, we propose a framework that allows the generator to utilize additional information from given measurements of training samples for prior learning, thereby yielding more accurate reconstruction for signals. As our framework has a simple form, it can be easily applied to existing CS methods using pre-trained generators. Through extensive experiments, we demonstrate that our framework consistently outperforms these works by a large margin and can reduce the reconstruction error up to an order of magnitude for the presented target applications. We also explain the experimental success theoretically by showing that our framework can slightly relax the stringent signal presence condition, which is required to guarantee the success of signal recovery.Kyung-Su KimJung Hyun LeeEunho YangIEEEarticleCompressed sensingartificial neural networksimage reconstructionimage enhancementsignal reconstruction and predictionmeasurement-conditional generative modelsElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 155335-155352 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Compressed sensing artificial neural networks image reconstruction image enhancement signal reconstruction and prediction measurement-conditional generative models Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Compressed sensing artificial neural networks image reconstruction image enhancement signal reconstruction and prediction measurement-conditional generative models Electrical engineering. Electronics. Nuclear engineering TK1-9971 Kyung-Su Kim Jung Hyun Lee Eunho Yang Compressed Sensing via Measurement-Conditional Generative Models
description	Pre-trained generators have been frequently adopted in compressed sensing (CS) owing to their ability to effectively estimate signals with the prior of NNs. To further refine the NN-based prior, we propose a framework that allows the generator to utilize additional information from given measurements of training samples for prior learning, thereby yielding more accurate reconstruction for signals. As our framework has a simple form, it can be easily applied to existing CS methods using pre-trained generators. Through extensive experiments, we demonstrate that our framework consistently outperforms these works by a large margin and can reduce the reconstruction error up to an order of magnitude for the presented target applications. We also explain the experimental success theoretically by showing that our framework can slightly relax the stringent signal presence condition, which is required to guarantee the success of signal recovery.
format	article
author	Kyung-Su Kim Jung Hyun Lee Eunho Yang
author_facet	Kyung-Su Kim Jung Hyun Lee Eunho Yang
author_sort	Kyung-Su Kim
title	Compressed Sensing via Measurement-Conditional Generative Models
title_short	Compressed Sensing via Measurement-Conditional Generative Models
title_full	Compressed Sensing via Measurement-Conditional Generative Models
title_fullStr	Compressed Sensing via Measurement-Conditional Generative Models
title_full_unstemmed	Compressed Sensing via Measurement-Conditional Generative Models
title_sort	compressed sensing via measurement-conditional generative models
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/bba1dc7b48244d7eade5d2db3dcb8225
work_keys_str_mv	AT kyungsukim compressedsensingviameasurementconditionalgenerativemodels AT junghyunlee compressedsensingviameasurementconditionalgenerativemodels AT eunhoyang compressedsensingviameasurementconditionalgenerativemodels
_version_	1718409991125204992

Compressed Sensing via Measurement-Conditional Generative Models

Ejemplares similares