FISTA-Net: Learning a Fast Iterative Shrinkage Thresholding Network for Inverse Problems in Imaging

Xiang, Jinxi; Dong, Yonggui; Yang, Yunjie

doi:10.1109/tmi.2021.3054167

Cited by 139 publications

(63 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Despite data can be segmented into patches to address this issue, segmented patches will cause the discontinuity of estimation results. Recently, some unfolded architecture of the iterative procedure is developed to form the framework of the deep learning network [27]- [30]. In [27], the architecture of unfolding FISTA is developed to form a deep network, where the parameters are learned through end-to-end training.…”

Section: A Recent Workmentioning

confidence: 99%

“…Recently, some unfolded architecture of the iterative procedure is developed to form the framework of the deep learning network [27]- [30]. In [27], the architecture of unfolding FISTA is developed to form a deep network, where the parameters are learned through end-to-end training. Subsequently, a new DL-based network architecture named alternating direction method of multipliers (ADMM)-Net was proposed, which was derived from the iterative procedures in the ADMM algorithm [28]- [30].…”

Section: A Recent Workmentioning

confidence: 99%

“…Inspired by [41], the optimization problem ( 27) can be solved by Wiener deconvolution technique. To learn a more suitable regularization kernel and parameters in (27), the Wiener filtering layer is developed as follows…”

Section: D) Convolution Layer (Cmentioning

confidence: 99%

See 2 more Smart Citations

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Chen

Tang

Zhang

et al. 2022

IEEE J. Sel. Top. Signal Process.

View full text Add to dashboard Cite

The emerging intelligent reflecting surface (IRS) can significantly improve the system capacity, and it has been regarded as a promising technology for the beyond fifth-generation (B5G) communications. For IRS-assisted multiple input multiple output (MIMO) systems, accurate channel estimation is a critical challenge. This severely restricts practical applications, particularly for resource-limited indoor scenario as it contains numerous scatterers and parameters to be estimated, while the number of pilots is limited. Prior art tackles these issues and associated optimization using mathematical-based statistical approaches, but are difficult to solve as the number of scatterers increase. To estimate the indoor channels with an affordable piloting overhead, we propose an offset learning (OL)-based neural network for channel estimation. The proposed OL-based estimator can dynamically trace the channel state information (CSI) without any prior knowledge of the IRS-assisted channel structure as well as indoor statistics. In addition, inspired by the powerful learning capability of convolutional neural network (CNN), CNN-based inversion blocks are developed in the offset estimation module to build the offset estimation operator. Numerical results show that the proposed OL-based estimator can achieve more accurate indoor

show abstract

Section: A Recent Workmentioning

confidence: 99%

Section: A Recent Workmentioning

confidence: 99%

See 1 more Smart Citation

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Chen

Tang

Zhang

et al. 2022

IEEE J. Sel. Top. Signal Process.

View full text Add to dashboard Cite

show abstract

“…That is also to say, gradient descent module v t computes the numerical solution with an adaptive step-size parameter τ t . As proved in [46], a good rule of thumb is to constrain τ t to be positive. Besides, the τ t should decay smoothly with iterations.…”

Section: ) Gradient Descent Module (V T )mentioning

confidence: 99%

“…As stated in [46], because of the progressively suppressed noise variances, the threshold value λ t should decrease with iterations accordingly. Thus, one defines a negative learnable weight w 2 , λ t can be constrained by (26) as well.…”

Section: ) Gradient Descent Module (V T )mentioning

confidence: 99%

Lightweight FISTA-Inspired Sparse Reconstruction Network for mmW 3-D Holography

Wang

Wei

Liang

et al. 2022

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

Integrating compressed sensing (CS) with millimeter-wave (mmW) holography has shown great potential to achieve lightweight onboard hardware, low sampling ratio, and high-speed sensing. However, conventional CS-driven algorithms are always limited by nontrivial adjusting of parameters and excessive computational cost caused by plenty of iterations. To address this problem, we propose a lightweight model-based deep learning framework (LFIST-Net) for mmW 3-D holography, by combining the interpretability of fast iterative shrinkage-thresholding algorithm (FISTA) and tuning-free merit of data-driven deep neural network. First, the single-frequency (SF) holographic imaging technique is integrated into FISTA, which serves as the sensing kernels, to avoid large-scale matrix multiplications. Subsequently, the kernel-based FISTA (KFISTA) is mapped into layer-fixed and parameter-learnable LFIST-Net, whose weights are relaxed to be layer-varied. The updating of key parameters in LFIST-Net, including step sizes, thresholds, and momentum coefficients, are regularized by soft-plus function to ensure the non-negativity and monotonicity. As for 3-D holography implementation, the "1-D + 2-D" scheme is adopted, where the matched filtering (MF) and well-trained LFIST-Net are used for range focusing and reconstructions of azimuth slices. Without losing efficiency, the range-focused subechoes are processed parallelly in 3-D cube form. Experiments, including both simulated and measured tests based on a commercial mmW radar, prove that LFIST-Net is capable of reconstructing the imaging scene precisely. In particular, in near-field mmW 3-D holography tests, both numerical and visual results demonstrate LFIST-Net yields compelling reconstruction performance while maintaining high computational speed compared with MF-based, conventional CS-driven, and network-based methods. Index Terms-3-D holography, compressed sensing (CS), deep unfolding, fast iterative shrinkage-thresholding algorithm (FISTA), millimeter-wave (mmW) imaging, near-field, sparse reconstruction. I. INTRODUCTIONN EAR-FIELD 3-D millimeter-wave (mmW) holography has gained significant concerns in the microwave theory and techniques society. mmW, possessing the merits of reasonable penetration for most nonmetal materials and no known Manuscript

show abstract

Image reconstruction based on a modified bird swarm optimization algorithm for electrical impedance tomography

Shi

Yuhang

Wang

et al. 2023

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Electrical impedance tomography (EIT) is a potential and promising tomographic technique. It is known that image reconstruction of EIT is mathematically an ill-posed inverse problem that leads to poor reconstruction quality. To improve the quality of reconstructed images, a novel image reconstruction method based on modified bird swarm optimization algorithm is proposed in this paper. An image of conductivity distribution acquired by Newton's onestep error reconstruction algorithm performs as initial population. Then a modified bird swarm optimization algorithm is proposed to optimize the initial conductivity distribution. By introducing a dynamic inertia weight parameter, flight interval of the bird swarm varies with the number of iterations. With the proposed method, a strong global search capability in the initial stage and a strong local search capability in the later stage are ensured during the optimization. To verify the effectiveness of the proposed method, simulation work and experimental validation are carried out. Also, reconstructed images are compared with the results obtained by iterative Landweber method and Newton-Raphson method. Meanwhile, the impact of noise on the reconstruction is investigated. The simulation and experimental results have demonstrated the superior performance of the proposed method in visualizing conductivity distribution.

show abstract

FISTA-Net: Learning a Fast Iterative Shrinkage Thresholding Network for Inverse Problems in Imaging

Cited by 139 publications

References 52 publications

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Offset Learning Based Channel Estimation for Intelligent Reflecting Surface-Assisted Indoor Communication

Lightweight FISTA-Inspired Sparse Reconstruction Network for mmW 3-D Holography

Image reconstruction based on a modified bird swarm optimization algorithm for electrical impedance tomography

Contact Info

Product

Resources

About