pith. sign in

arxiv: 2301.06081 · v4 · submitted 2022-12-09 · 📡 eess.IV · cs.CV

A Data-driven Loss Weighting Scheme across Heterogeneous Tasks for Image Denoising

Xiangyu Rui , Xiangyong Cao , Xile Zhao , Deyu Meng , Michael K. Ng This is my paper

Pith reviewed 2026-05-24 10:00 UTC · model grok-4.3

classification 📡 eess.IV cs.CV
keywords image denoisingvariational modelsloss weightingbilevel optimizationdata-driven weightingcomplex noisetransfer learninggeneralization
0
0 comments X

The pith

A neural network trained by bilevel optimization predicts weights for the data fidelity term that improve variational denoising on complex noise and transfer across models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a data-driven loss weighting scheme where a neural network maps a noisy image to a weight value used in the fidelity term of variational denoising models. Training occurs through bilevel optimization: the lower level solves multiple denoising problems that share the same predicted weight, while the upper level minimizes the gap between the restored image and the clean ground truth. This joint extraction of noise and regularization information yields a weight function that can be plugged into different variational models. Experiments show the approach handles impulse noise, stripe noise, and mixtures better than standard weighting, and the same function works on tasks not seen during training.

Core claim

The central claim is that training a parameterized weight function (neural network) mapping noisy images to weights using bilevel optimization enables variational denoising models to better handle complex noise patterns like impulse or stripe noise, and the learned weights transfer to other models and tasks beyond training.

What carries the argument

The DLW weight function, a neural network trained by bilevel optimization to output data-fidelity weights from noisy images.

If this is right

  • Variational denoising models gain the ability to process impulse noise, stripe noise, and mixed patterns without manual weight tuning.
  • A single trained weight function can be inserted into multiple different regularization-based denoisers.
  • Noise-handling knowledge learned at the model level transfers to tasks outside the original training distribution.
  • Generalization bounds support that the learned weighting is intrinsically transferable rather than tied to one specific noise model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same bilevel construction could be applied to other inverse problems where the balance between data term and prior must adapt to unknown degradations.
  • Once trained, the weight network might serve as a plug-in module for non-variational denoisers that still contain an explicit fidelity term.
  • Testing on real sensor data with spatially varying noise would check whether the learned mapping generalizes beyond the synthetic patterns used in training.

Load-bearing premise

A weight function trained on one set of noise patterns and regularization terms will still produce useful weights when the noise or the regularizer changes.

What would settle it

Retraining the weight function on one noise type and then measuring whether a variational model using that function on a held-out noise type performs worse than the same model with a fixed or hand-tuned weight.

read the original abstract

In a variational denoising model, weight in the data fidelity term plays the role of enhancing the noise-removal capability. It is profoundly correlated with noise information, while also balancing the data fidelity and regularization terms. However, the difficulty of assigning weight is expected to be substantial when the noise pattern is beyond independent identical Gaussian distribution, e.g., impulse noise, stripe noise, or a mixture of several patterns, etc. Furthermore, how to leverage weight to balance the data fidelity and regularization terms is even less evident. In this work, we propose a data-driven loss weighting (DLW) scheme to address these issues. Specifically, DLW trains a parameterized weight function (i.e., a neural network) that maps the noisy image to the weight. The training is achieved by a bilevel optimization framework, where the lower level problem is solving several denoising models with the same weight predicted by the weight function and the upper level problem minimizes the distance between the restored image and the clean image. In this way, information from both the noise and the regularization can be efficiently extracted to determine the weight function. DLW also facilitates the easy implementation of a trained weight function on denoising models. Numerical results verify the remarkable performance of DLW on improving the ability of various variational denoising models to handle different complex noise. This implies that DLW has the ability to transfer the noise knowledge at the model level to heterogeneous tasks beyond the training ones and the generalization theory underlying DLW is studied, validating its intrinsic transferability.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes a data-driven loss weighting (DLW) scheme for variational image denoising. A neural network is trained via bilevel optimization to predict the weight in the data-fidelity term: the lower level solves several denoising models that share the same predicted weight, while the upper level minimizes the distance between the restored and clean images. The central claims are that DLW improves performance on complex (non-i.i.d. Gaussian) noise patterns and that the learned weight function transfers to heterogeneous tasks beyond those seen in training, with supporting numerical results and a generalization theory.

Significance. If the transfer claim holds, the method would offer a practical way to obtain adaptive, model-agnostic fidelity weights that incorporate both noise statistics and regularization effects without manual tuning. The bilevel construction that extracts information from multiple regularizers simultaneously is a clear technical contribution; the explicit study of generalization theory is also a strength.

major comments (2)
  1. [Abstract] Abstract (and § on numerical results): the claim that DLW 'has the ability to transfer the noise knowledge at the model level to heterogeneous tasks beyond the training ones' is load-bearing. The manuscript must explicitly state whether the regularization terms used in the test models were excluded from the lower-level collection during upper-level training; without this, the reported improvements on 'various variational denoising models' do not demonstrate the required out-of-distribution behavior.
  2. [Generalization theory] Generalization theory section: the theory is invoked to 'validate its intrinsic transferability,' yet the bilevel objective is defined only with respect to the specific regularizers present in the lower level. If the theory does not derive an invariance or bound with respect to a change in the regularization operator, the transfer claim rests on an unproven assumption.
minor comments (1)
  1. [Method] The abstract states that the weight 'is profoundly correlated with noise information' but does not clarify whether the network input is the noisy image alone or also includes noise-level estimates; this notation should be made explicit in the method section.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the scope of our transferability claims. We address each major point below and will revise the manuscript accordingly to improve precision without altering the core contributions.

read point-by-point responses

  1. Referee: [Abstract] Abstract (and § on numerical results): the claim that DLW 'has the ability to transfer the noise knowledge at the model level to heterogeneous tasks beyond the training ones' is load-bearing. The manuscript must explicitly state whether the regularization terms used in the test models were excluded from the lower-level collection during upper-level training; without this, the reported improvements on 'various variational denoising models' do not demonstrate the required out-of-distribution behavior.

    Authors: We confirm that the regularization terms appearing in the test models were excluded from the lower-level collection used during upper-level training; this was done precisely to evaluate out-of-distribution transfer. The current manuscript does not state this exclusion explicitly, which we agree weakens the presentation of the transfer claim. We will revise both the abstract and the numerical-results section to include a clear statement of the experimental protocol, thereby documenting the out-of-distribution nature of the reported improvements. revision: yes

  2. Referee: [Generalization theory] Generalization theory section: the theory is invoked to 'validate its intrinsic transferability,' yet the bilevel objective is defined only with respect to the specific regularizers present in the lower level. If the theory does not derive an invariance or bound with respect to a change in the regularization operator, the transfer claim rests on an unproven assumption.

    Authors: The generalization bounds are derived under the bilevel formulation that trains a single weight function shared across multiple regularizers; they therefore quantify stability with respect to the training distribution of regularizers. The theory does not, however, establish an explicit invariance or bound for an arbitrary unseen regularization operator. We will revise the theory section to state this scope limitation explicitly and to clarify that transfer to completely novel regularizers is supported primarily by the numerical evidence rather than by the current theoretical guarantees. revision: yes

Circularity Check

0 steps flagged

No circularity detected in bilevel training or generalization claims.

full rationale

The bilevel optimization trains the weight function (neural network) by using external clean images in the upper-level objective to minimize reconstruction error from lower-level variational models that share the predicted weight; this does not reduce any claimed prediction to a quantity defined by the weight function itself. No equations or text in the provided abstract equate the learned mapping to its own outputs by construction, rename a fitted parameter as a prediction, or rely on self-citations for load-bearing uniqueness or ansatz. The generalization theory is described as an independent study, and numerical results on various models constitute external validation rather than a self-referential loop. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the existence of a solvable lower-level denoising problem for any predicted weight and on the availability of paired clean-noisy training data for the upper level. The neural-network parameters constitute the only fitted quantities.

free parameters (1)
  • neural network parameters of the weight function
    Learned during upper-level optimization to map noisy images to weights.
axioms (1)
  • domain assumption The variational denoising problem admits a solution for any fixed weight produced by the network.
    Invoked when the lower-level problem is solved inside the bilevel loop.

pith-pipeline@v0.9.0 · 5813 in / 1133 out tokens · 39988 ms · 2026-05-24T10:00:43.507894+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

85 extracted references · 85 canonical work pages

  1. [1]

    IEEE Geoscience and Remote Sensing Letters 13(3):442--446

    Aggarwal HK, Majumdar A (2016) Hyperspectral image denoising using spatio-spectral total variation. IEEE Geoscience and Remote Sensing Letters 13(3):442--446

    work page 2016

  2. [2]

    In: European Conference on Computer Vision, Springer, pp 19--34

    Arad B, Ben-Shahar O (2016) Sparse recovery of hyperspectral signal from natural rgb images. In: European Conference on Computer Vision, Springer, pp 19--34

    work page 2016

  3. [3]

    Journal of Machine Learning Research 3(Nov):463--482

    Bartlett PL, Mendelson S (2002) Rademacher and gaussian complexities: Risk bounds and structural results. Journal of Machine Learning Research 3(Nov):463--482

    work page 2002

  4. [4]

    Society for Industrial and Applied Mathematics

    Beck A (2017) First-Order Methods in Optimization. Society for Industrial and Applied Mathematics

    work page 2017

  5. [5]

    Machine Learning 79(1):151--175

    Ben-David S, Blitzer J, Crammer K, et al (2010) A theory of learning from different domains. Machine Learning 79(1):151--175

    work page 2010

  6. [6]

    Advances in Neural Information Processing Systems 24

    Blanchard G, Lee G, Scott C (2011) Generalizing from several related classification tasks to a new unlabeled sample. Advances in Neural Information Processing Systems 24

    work page 2011

  7. [7]

    The Journal of Machine Learning Research 22(1):46--100

    Blanchard G, Deshmukh AA, Dogan \"U , et al (2021) Domain generalization by marginal transfer learning. The Journal of Machine Learning Research 22(1):46--100

    work page 2021

  8. [8]

    Advances in Neural Information Processing Systems 34:5430--5442

    Bodrito T, Zouaoui A, Chanussot J, et al (2021) A trainable spectral-spatial sparse coding model for hyperspectral image restoration. Advances in Neural Information Processing Systems 34:5430--5442

    work page 2021

  9. [9]

    Foundations and Trends in Machine Learning 3(1):1--122

    Boyd S, Parikh N, Chu E, et al (2011) Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends in Machine Learning 3(1):1--122

    work page 2011

  10. [10]

    Journal of Fourier Analysis and Applications 14(5):877--905

    Candes EJ, Wakin MB, Boyd SP (2008) Enhancing sparsity by reweighted _1 minimization. Journal of Fourier Analysis and Applications 14(5):877--905

    work page 2008

  11. [11]

    IEEE Transactions on Image Processing 25(10):4677--4690

    Cao X, Zhao Q, Meng D, et al (2016) Robust low-rank matrix factorization under general mixture noise distributions. IEEE Transactions on Image Processing 25(10):4677--4690

    work page 2016

  12. [12]

    IEEE Transactions on Geoscience and Remote Sensing 60:1--14

    Cao X, Fu X, Xu C, et al (2021) Deep spatial-spectral global reasoning network for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 60:1--14

    work page 2021

  13. [13]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 4260--4268

    Chang Y, Yan L, Zhong S (2017) Hyper-laplacian regularized unidirectional low-rank tensor recovery for multispectral image denoising. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 4260--4268

    work page 2017

  14. [14]

    IEEE Transactions on Geoscience and Remote Sensing 57(2):667--682

    Chang Y, Yan L, Fang H, et al (2018) Hsi-denet: Hyperspectral image restoration via convolutional neural network. IEEE Transactions on Geoscience and Remote Sensing 57(2):667--682

    work page 2018

  15. [15]

    IEEE Transactions on Cybernetics 48(3):1054--1066

    Chen Y, Cao X, Zhao Q, et al (2017 a ) Denoising hyperspectral image with non-iid noise structure. IEEE Transactions on Cybernetics 48(3):1054--1066

    work page 2017

  16. [16]

    IEEE Transactions on Geoscience and Remote Sensing 55(9):5366--5380

    Chen Y, Guo Y, Wang Y, et al (2017 b ) Denoising of hyperspectral images using nonconvex low rank matrix approximation. IEEE Transactions on Geoscience and Remote Sensing 55(9):5366--5380

    work page 2017

  17. [17]

    IEEE Transactions on Geoscience and Remote Sensing 58(2):1348--1362

    Chen Y, He W, Yokoya N, et al (2019) Nonlocal tensor-ring decomposition for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 58(2):1348--1362

    work page 2019

  18. [18]

    In: International Conference on Algorithmic Learning Theory, Springer, pp 308--323

    Cortes C, Mohri M (2011) Domain adaptation in regression. In: International Conference on Algorithmic Learning Theory, Springer, pp 308--323

    work page 2011

  19. [19]

    In: International Conference on Machine Learning, pp 193--200

    Dai Wenyuan YQ, Guirong X, Yong Y (2007) Boosting for transfer learning. In: International Conference on Machine Learning, pp 193--200

    work page 2007

  20. [20]

    In: International Conference on Machine Learning, PMLR, pp 859--868

    Germain P, Habrard A, Laviolette F, et al (2016) A new pac-bayesian perspective on domain adaptation. In: International Conference on Machine Learning, PMLR, pp 859--868

    work page 2016

  21. [21]

    International Journal of Computer Vision 121(2):183--208

    Gu S, Xie Q, Meng D, et al (2017) Weighted nuclear norm minimization and its applications to low level vision. International Journal of Computer Vision 121(2):183--208

    work page 2017

  22. [22]

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 8(6):3050--3061

    He W, Zhang H, Zhang L, et al (2015 a ) Hyperspectral image denoising via noise-adjusted iterative low-rank matrix approximation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 8(6):3050--3061

    work page 2015

  23. [23]

    IEEE Transactions on Geoscience and Remote Sensing 54(1):178--188

    He W, Zhang H, Zhang L, et al (2015 b ) Total-variation-regularized low-rank matrix factorization for hyperspectral image restoration. IEEE Transactions on Geoscience and Remote Sensing 54(1):178--188

    work page 2015

  24. [24]

    IEEE Transactions on Pattern Analysis and Machine Intelligence

    He W, Yao Q, Li C, et al (2020) Non-local meets global: An integrated paradigm for hyperspectral image restoration. IEEE Transactions on Pattern Analysis and Machine Intelligence

    work page 2020

  25. [25]

    Canadian Journal of Remote Sensing 42(1):53--72

    Jiang C, Zhang H, Zhang L, et al (2016) Hyperspectral image denoising with a combined spatial and spectral weighted hyperspectral total variation model. Canadian Journal of Remote Sensing 42(1):53--72

    work page 2016

  26. [26]

    IEEE Transactions on Geoscience and Remote Sensing 60:1--13

    Jiang TX, Zhuang L, Huang TZ, et al (2021) Adaptive hyperspectral mixed noise removal. IEEE Transactions on Geoscience and Remote Sensing 60:1--13

    work page 2021

  27. [27]

    In: Imaging Spectrometry III, SPIE, pp 57--68

    Kalman LS, Bassett III EM (1997) Classification and material identification in an urban environment using hydice hyperspectral data. In: Imaging Spectrometry III, SPIE, pp 57--68

    work page 1997

  28. [28]

    arXiv preprint arXiv:14126980

    Kingma DP, Ba J (2014) Adam: A method for stochastic optimization. arXiv preprint arXiv:14126980

    work page 2014

  29. [29]

    In: International Conference on Machine Learning, PMLR, pp 2965--2974

    Lehtinen J, Munkberg J, Hasselgren J, et al (2018) Noise2noise: Learning image restoration without clean data. In: International Conference on Machine Learning, PMLR, pp 2965--2974

    work page 2018

  30. [30]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, pp 1368--1376

    Li M, Fu Y, Zhang Y (2023 a ) Spatial-spectral transformer for hyperspectral image denoising. In: Proceedings of the AAAI Conference on Artificial Intelligence, pp 1368--1376

    work page 2023

  31. [31]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp 5805--5814

    Li M, Liu J, Fu Y, et al (2023 b ) Spectral enhanced rectangle transformer for hyperspectral image denoising. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp 5805--5814

    work page 2023

  32. [32]

    IEEE Transactions on Image Processing 29:565--578

    Lin B, Tao X, Lu J (2019) Hyperspectral image denoising via matrix factorization and deep prior regularization. IEEE Transactions on Image Processing 29:565--578

    work page 2019

  33. [33]

    IEEE Transactions on Geoscience and Remote Sensing 59(9):7739--7757

    Lin J, Huang TZ, Zhao XL, et al (2020) A tensor subspace representation-based method for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 59(9):7739--7757

    work page 2020

  34. [34]

    Food and Bioprocess Technology 7(2):307--323

    Liu D, Sun DW, Zeng XA (2014) Recent advances in wavelength selection techniques for hyperspectral image processing in the food industry. Food and Bioprocess Technology 7(2):307--323

    work page 2014

  35. [35]

    Remote Sensing 12(16):2659

    Lu B, Dao PD, Liu J, et al (2020) Recent advances of hyperspectral imaging technology and applications in agriculture. Remote Sensing 12(16):2659

    work page 2020

  36. [36]

    IEEE Transactions on Geoscience and Remote Sensing 54(1):373--385

    Lu T, Li S, Fang L, et al (2015) Spectral--spatial adaptive sparse representation for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 54(1):373--385

    work page 2015

  37. [37]

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 14:9435--9449

    Luo YS, Zhao XL, Jiang TX, et al (2021) Hyperspectral mixed noise removal via spatial-spectral constrained unsupervised deep image prior. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 14:9435--9449

    work page 2021

  38. [38]

    IEEE Transactions on Geoscience and Remote Sensing 55(3):1227--1239

    Ma Y, Li C, Mei X, et al (2016) Robust sparse hyperspectral unmixing with _ 2, 1 norm. IEEE Transactions on Geoscience and Remote Sensing 55(3):1227--1239

    work page 2016

  39. [39]

    Advances in Neural Information Processing Systems 20

    Mahmud M, Ray S (2007) Transfer learning using kolmogorov complexity: Basic theory and empirical evaluations. Advances in Neural Information Processing Systems 20

    work page 2007

  40. [40]

    Cambridge University Press

    Manolakis DG, Lockwood RB, Cooley TW (2016) Hyperspectral imaging remote sensing: physics, sensors, and algorithms. Cambridge University Press

    work page 2016

  41. [41]

    In: International Conference on Algorithmic Learning Theory, Springer, pp 3--17

    Maurer A (2016) A vector-contraction inequality for rademacher complexities. In: International Conference on Algorithmic Learning Theory, Springer, pp 3--17

    work page 2016

  42. [42]

    Journal of Machine Learning Research 17(81):1--32

    Maurer A, Pontil M, Romera-Paredes B (2016) The benefit of multitask representation learning. Journal of Machine Learning Research 17(81):1--32

    work page 2016

  43. [43]

    In: International Conference on Machine Learning, PMLR, pp 2373--2381

    McNamara D, Balcan MF (2017) Risk bounds for transferring representations with and without fine-tuning. In: International Conference on Machine Learning, PMLR, pp 2373--2381

    work page 2017

  44. [44]

    In: Proceedings of the IEEE International Conference on Computer Vision, pp 1337--1344

    Meng D, De La Torre F (2013) Robust matrix factorization with unknown noise. In: Proceedings of the IEEE International Conference on Computer Vision, pp 1337--1344

    work page 2013

  45. [45]

    MIT press

    Mohri M, Rostamizadeh A, Talwalkar A (2018) Foundations of machine learning. MIT press

    work page 2018

  46. [46]

    In: International Conference on Machine Learning, PMLR, pp 10--18

    Muandet K, Balduzzi D, Sch \"o lkopf B (2013) Domain generalization via invariant feature representation. In: International Conference on Machine Learning, PMLR, pp 10--18

    work page 2013

  47. [47]

    Remote Sensing 14(18)

    Pang L, Gu W, Cao X (2022) Trq3dnet: A 3d quasi-recurrent and transformer based network for hyperspectral image denoising. Remote Sensing 14(18)

    work page 2022

  48. [48]

    Foundations and Trends in Optimization 1(3):127--239

    Parikh N, Boyd S (2014) Proximal algorithms. Foundations and Trends in Optimization 1(3):127--239

    work page 2014

  49. [49]

    IEEE Transactions on Image Processing 29:7889--7903

    Peng J, Xie Q, Zhao Q, et al (2020) Enhanced 3dtv regularization and its applications on hsi denoising and compressed sensing. IEEE Transactions on Image Processing 29:7889--7903

    work page 2020

  50. [50]

    IEEE Transactions on Geoscience and Remote Sensing 60:1--17

    Peng J, Wang H, Cao X, et al (2022 a ) Fast noise removal in hyperspectral images via representative coefficient total variation. IEEE Transactions on Geoscience and Remote Sensing 60:1--17

    work page 2022

  51. [51]

    IEEE Transactions on Pattern Analysis and Machine Intelligence pp 1--16

    Peng J, Wang Y, Zhang H, et al (2022 b ) Exact decomposition of joint low rankness and local smoothness plus sparse matrices. IEEE Transactions on Pattern Analysis and Machine Intelligence pp 1--16

    work page 2022

  52. [52]

    Remote Sensing of Environment 233:111350

    Pyo J, Duan H, Baek S, et al (2019) A convolutional neural network regression for quantifying cyanobacteria using hyperspectral imagery. Remote Sensing of Environment 233:111350

    work page 2019

  53. [53]

    In: Proceedings of the IEEE International Conference on Computer Vision, pp 5533--5541

    Qiu Z, Yao T, Mei T (2017) Learning spatio-temporal representation with pseudo-3d residual networks. In: Proceedings of the IEEE International Conference on Computer Vision, pp 5533--5541

    work page 2017

  54. [54]

    IEEE Geoscience and Remote Sensing Letters 14(12):2335--2339

    Rasti B, Ulfarsson MO, Ghamisi P (2017) Automatic hyperspectral image restoration using sparse and low-rank modeling. IEEE Geoscience and Remote Sensing Letters 14(12):2335--2339

    work page 2017

  55. [55]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 6739--6748

    Rui X, Cao X, Xie Q, et al (2021) Learning an explicit weighting scheme for adapting complex hsi noise. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 6739--6748

    work page 2021

  56. [56]

    IEEE Transactions on Geoscience and Remote Sensing 59(12):10348--10363

    Shi Q, Tang X, Yang T, et al (2021) Hyperspectral image denoising using a 3-d attention denoising network. IEEE Transactions on Geoscience and Remote Sensing 59(12):10348--10363

    work page 2021

  57. [57]

    IEEE Transactions on Signal Processing 63(22):6013--6023

    Shi W, Ling Q, Wu G, et al (2015) A proximal gradient algorithm for decentralized composite optimization. IEEE Transactions on Signal Processing 63(22):6013--6023

    work page 2015

  58. [58]

    arXiv preprint arXiv:210702378

    Shu J, Meng D, Xu Z (2021) Learning an explicit hyperparameter prediction policy conditioned on tasks. arXiv preprint arXiv:210702378

    work page 2021

  59. [59]

    arXiv preprint arXiv:210203924

    Sicilia A, Zhao X, Hwang SJ (2021) Domain adversarial neural networks for domain generalization: When it works and how to improve. arXiv preprint arXiv:210203924

    work page 2021

  60. [60]

    IEEE Transactions on Multimedia 22(12):3236--3248

    Sun Y, Yang Y, Liu Q, et al (2020) Learning non-locally regularized compressed sensing network with half-quadratic splitting. IEEE Transactions on Multimedia 22(12):3236--3248

    work page 2020

  61. [61]

    Advances in Neural Information Processing Systems 33:7852--7862

    Tripuraneni N, Jordan M, Jin C (2020) On the theory of transfer learning: The importance of task diversity. Advances in Neural Information Processing Systems 33:7852--7862

    work page 2020

  62. [62]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 9446--9454

    Ulyanov D, Vedaldi A, Lempitsky V (2018) Deep image prior. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp 9446--9454

    work page 2018

  63. [63]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp 0--0

    Wang P (2019) Image denoising using deep cgan with bi-skip connections. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp 0--0

    work page 2019

  64. [64]

    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 11(4):1227--1243

    Wang Y, Peng J, Zhao Q, et al (2017) Hyperspectral image restoration via total variation regularized low-rank tensor decomposition. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 11(4):1227--1243

    work page 2017

  65. [65]

    IEEE Transactions on Neural Networks and Learning Systems 32(1):363--375

    Wei K, Fu Y, Huang H (2020) 3-d quasi-recurrent neural network for hyperspectral image denoising. IEEE Transactions on Neural Networks and Learning Systems 32(1):363--375

    work page 2020

  66. [66]

    IEEE Transactions on Geoscience and Remote Sensing 55(12):6860--6876

    Wei W, Zhang L, Tian C, et al (2017) Structured sparse coding-based hyperspectral imagery denoising with intracluster filtering. IEEE Transactions on Geoscience and Remote Sensing 55(12):6860--6876

    work page 2017

  67. [67]

    IEEE Transactions on Pattern Analysis and Machine Intelligence 40(8):1888--1902

    Xie Q, Zhao Q, Meng D, et al (2017) Kronecker-basis-representation based tensor sparsity and its applications to tensor recovery. IEEE Transactions on Pattern Analysis and Machine Intelligence 40(8):1888--1902

    work page 2017

  68. [68]

    IEEE Transactions on Pattern Analysis and Machine Intelligence pp 1--14

    Xie X, Wang Q, Ling Z, et al (2022) Optimization induced equilibrium networks: An explicit optimization perspective for understanding equilibrium models. IEEE Transactions on Pattern Analysis and Machine Intelligence pp 1--14

    work page 2022

  69. [69]

    arXiv preprint arXiv:201201829

    Xiong F, Tao S, Zhou J, et al (2020) Smds-net: Model guided spectral-spatial network for hyperspectral image denoising. arXiv preprint arXiv:201201829

    work page 2020

  70. [70]

    IEEE Transactions on Geoscience and Remote Sensing 60:1--14

    Xiong F, Zhou J, Zhao Q, et al (2021) Mac-net: Model-aided nonlocal neural network for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 60:1--14

    work page 2021

  71. [71]

    IEEE Transactions on Geoscience and Remote Sensing 57(7):5174--5189

    Xue J, Zhao Y, Liao W, et al (2019) Nonlocal low-rank regularized tensor decomposition for hyperspectral image denoising. IEEE Transactions on Geoscience and Remote Sensing 57(7):5174--5189

    work page 2019

  72. [72]

    IEEE transactions on image processing 19(9):2241--2253

    Yasuma F, Mitsunaga T, Iso D, et al (2010) Generalized assorted pixel camera: postcapture control of resolution, dynamic range, and spectrum. IEEE transactions on image processing 19(9):2241--2253

    work page 2010

  73. [73]

    IEEE Transactions on Geoscience and Remote Sensing 50(10):3660--3677

    Yuan Q, Zhang L, Shen H (2012) Hyperspectral image denoising employing a spectral--spatial adaptive total variation model. IEEE Transactions on Geoscience and Remote Sensing 50(10):3660--3677

    work page 2012

  74. [74]

    IEEE Transactions on Geoscience and Remote Sensing 57(2):1205--1218

    Yuan Q, Zhang Q, Li J, et al (2018) Hyperspectral image denoising employing a spatial--spectral deep residual convolutional neural network. IEEE Transactions on Geoscience and Remote Sensing 57(2):1205--1218

    work page 2018

  75. [75]

    Remote Sensing 10(10):1631

    Yue Z, Meng D, Sun Y, et al (2018) Hyperspectral image restoration under complex multi-band noises. Remote Sensing 10(10):1631

    work page 2018

  76. [76]

    IEEE Transactions on Geoscience and Remote Sensing 52(8):4729--4743

    Zhang H, He W, Zhang L, et al (2013) Hyperspectral image restoration using low-rank matrix recovery. IEEE Transactions on Geoscience and Remote Sensing 52(8):4729--4743

    work page 2013

  77. [77]

    IEEE Transactions on Geoscience and Remote Sensing 58(5):3071--3084

    Zhang H, Liu L, He W, et al (2019) Hyperspectral image denoising with total variation regularization and nonlocal low-rank tensor decomposition. IEEE Transactions on Geoscience and Remote Sensing 58(5):3071--3084

    work page 2019

  78. [78]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp 2248--2257

    Zhang T, Fu Y, Li C (2021) Hyperspectral image denoising with realistic data. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp 2248--2257

    work page 2021

  79. [79]

    International Journal of Computer Vision 130(11):2885--2901

    Zhang T, Fu Y, Zhang J (2022) Guided hyperspectral image denoising with realistic data. International Journal of Computer Vision 130(11):2885--2901

    work page 2022

  80. [80]

    IEEE Transactions on Geoscience and Remote Sensing 53(1):296--308

    Zhao YQ, Yang J (2014) Hyperspectral image denoising via sparse representation and low-rank constraint. IEEE Transactions on Geoscience and Remote Sensing 53(1):296--308

    work page 2014

Showing first 80 references.