Deep learning is playing an increasingly important role in signal processing field due to its excellent performance on many inference problems. Parametric bilinear generalized approximate message passing (P-BiG-AMP) is a new approximate message passing based approach to a general class of structure-matrix bilinear estimation problems. In this letter, we propose a novel feed-forward neural network architecture to realize P-BiG-AMP methodology with deep learning for the inference problem of compressive sensing under matrix uncertainty. Linear transforms utilized in the recovery process and parameters involved in the input and output channels of measurement are jointly learned from training data. Simulation results show that the trained P-BiG-AMP network can achieve higher reconstruction performance than the P-BiG-AMP algorithm with parameters tuned via the expectation-maximization method.
Jingjing SI
Yanshan University,Hebei Key Laboratory of Information Transmission and Signal Processing
Wenwen SUN
Yanshan University
Chuang LI
Yanshan University
Yinbo CHENG
Hebei Agricultural University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Jingjing SI, Wenwen SUN, Chuang LI, Yinbo CHENG, "Deep Network for Parametric Bilinear Generalized Approximate Message Passing and Its Application in Compressive Sensing under Matrix Uncertainty" in IEICE TRANSACTIONS on Fundamentals,
vol. E104-A, no. 4, pp. 751-756, April 2021, doi: 10.1587/transfun.2020EAL2050.
Abstract: Deep learning is playing an increasingly important role in signal processing field due to its excellent performance on many inference problems. Parametric bilinear generalized approximate message passing (P-BiG-AMP) is a new approximate message passing based approach to a general class of structure-matrix bilinear estimation problems. In this letter, we propose a novel feed-forward neural network architecture to realize P-BiG-AMP methodology with deep learning for the inference problem of compressive sensing under matrix uncertainty. Linear transforms utilized in the recovery process and parameters involved in the input and output channels of measurement are jointly learned from training data. Simulation results show that the trained P-BiG-AMP network can achieve higher reconstruction performance than the P-BiG-AMP algorithm with parameters tuned via the expectation-maximization method.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2020EAL2050/_p
Copy
@ARTICLE{e104-a_4_751,
author={Jingjing SI, Wenwen SUN, Chuang LI, Yinbo CHENG, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Deep Network for Parametric Bilinear Generalized Approximate Message Passing and Its Application in Compressive Sensing under Matrix Uncertainty},
year={2021},
volume={E104-A},
number={4},
pages={751-756},
abstract={Deep learning is playing an increasingly important role in signal processing field due to its excellent performance on many inference problems. Parametric bilinear generalized approximate message passing (P-BiG-AMP) is a new approximate message passing based approach to a general class of structure-matrix bilinear estimation problems. In this letter, we propose a novel feed-forward neural network architecture to realize P-BiG-AMP methodology with deep learning for the inference problem of compressive sensing under matrix uncertainty. Linear transforms utilized in the recovery process and parameters involved in the input and output channels of measurement are jointly learned from training data. Simulation results show that the trained P-BiG-AMP network can achieve higher reconstruction performance than the P-BiG-AMP algorithm with parameters tuned via the expectation-maximization method.},
keywords={},
doi={10.1587/transfun.2020EAL2050},
ISSN={1745-1337},
month={April},}
Copy
TY - JOUR
TI - Deep Network for Parametric Bilinear Generalized Approximate Message Passing and Its Application in Compressive Sensing under Matrix Uncertainty
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 751
EP - 756
AU - Jingjing SI
AU - Wenwen SUN
AU - Chuang LI
AU - Yinbo CHENG
PY - 2021
DO - 10.1587/transfun.2020EAL2050
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E104-A
IS - 4
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - April 2021
AB - Deep learning is playing an increasingly important role in signal processing field due to its excellent performance on many inference problems. Parametric bilinear generalized approximate message passing (P-BiG-AMP) is a new approximate message passing based approach to a general class of structure-matrix bilinear estimation problems. In this letter, we propose a novel feed-forward neural network architecture to realize P-BiG-AMP methodology with deep learning for the inference problem of compressive sensing under matrix uncertainty. Linear transforms utilized in the recovery process and parameters involved in the input and output channels of measurement are jointly learned from training data. Simulation results show that the trained P-BiG-AMP network can achieve higher reconstruction performance than the P-BiG-AMP algorithm with parameters tuned via the expectation-maximization method.
ER -