Wendong Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Wendong Wang has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Jianjun Wang
Southwest University
18 shared publications - Feng Zhang
School of Mathematics and Statistics, Southwest University, Chongqing, China
9 shared publications - Jianwen Huang
School of Mathematical Sciences, Chongqing Normal University, Chongqing, China
6 shared publications - Zhi Wang
College of Computer and Information Science, Southwest University, Chongqing, China
3 shared publications Feng Zhang
2 shared publicationsJianjun Wang
Southwest University, College of Artificial Intelligence, Chongqing, China
2 shared publications
Latest Publications
- 7 December 2025
- Published by Elsevier in Signal Processing
- Vol. 242
New sufficient condition of ℓ 1 − β ℓ q method for robust signal recovery
- 7 December 2025
- Published by Elsevier in Signal Processing
- Vol. 242
New sufficient condition of ℓ 1 − β ℓ q method for robust signal recovery
- 1 May 2025
- Published by SPIE-Intl Soc Optical Eng in Journal of Electronic Imaging
- Vol. 34 (03) , 033016-17
Outlier-robust image inpainting via nuclear norm-guided bilinear factorization
Matrix factorization (MF)-based methods are very popular in image inpainting fields. However, the existing MF-based methods are mainly considered in the noiseless case or the cases where the noise is of Gaussian type. Although there exist a few MF-based methods that were proposed to resist the outliers and impulsive noise, they do not well pose some effective constraints on the factor matrices, which to some extent limits the resultant restoration performance. Inspired by a HOAT (half-quadratic optimization based on additive form for truncated quadratic function) algorithm proposed recently, we first consider adding a nuclear norm-guided bilinear factorization constraint into it, and based on the popular block coordinate descent framework; thus, we propose an improved HOAT algorithm to deal with the outlier-robust image inpainting tasks. Surprisingly, this small modification makes our algorithm possess the capability to perform much better than the HOAT algorithm in both restoration accuracy and running time when the rank estimate tends to be large. By contrast, the HOAT algorithm will result in much worse performance under the same settings of rank estimate. Extensive experiments on outlier-robust image inpainting with random/non-random masks further verify the effectiveness of the proposed algorithm when it is compared with some state-of-the-art MF-based algorithms.
- 1 May 2025
- Published by SPIE-Intl Soc Optical Eng in Journal of Electronic Imaging
- Vol. 34 (03) , 033016-17
Outlier-robust image inpainting via nuclear norm-guided bilinear factorization
Matrix factorization (MF)-based methods are very popular in image inpainting fields. However, the existing MF-based methods are mainly considered in the noiseless case or the cases where the noise is of Gaussian type. Although there exist a few MF-based methods that were proposed to resist the outliers and impulsive noise, they do not well pose some effective constraints on the factor matrices, which to some extent limits the resultant restoration performance. Inspired by a HOAT (half-quadratic optimization based on additive form for truncated quadratic function) algorithm proposed recently, we first consider adding a nuclear norm-guided bilinear factorization constraint into it, and based on the popular block coordinate descent framework; thus, we propose an improved HOAT algorithm to deal with the outlier-robust image inpainting tasks. Surprisingly, this small modification makes our algorithm possess the capability to perform much better than the HOAT algorithm in both restoration accuracy and running time when the rank estimate tends to be large. By contrast, the HOAT algorithm will result in much worse performance under the same settings of rank estimate. Extensive experiments on outlier-robust image inpainting with random/non-random masks further verify the effectiveness of the proposed algorithm when it is compared with some state-of-the-art MF-based algorithms.
- 1 September 2024
- Published by SPIE-Intl Soc Optical Eng in Journal of Electronic Imaging
- Vol. 33 (05) , 053062-18
Exact matrix completion via smooth matrix factorization
Matrix completion (MC) has been successfully applied to many real-world applications, such as recommender systems and image inpainting. Traditional MC methods mainly aim at minimizing the rank function or its surrogate functions. However, the resulting algorithms inevitably calculate the singular value decomposition (SVD) at each iteration, which not only consumes a great deal of time but also greatly limits the potential applications in large-scale data. We propose an MC method that is based on smooth matrix factorization (MF) to deal with the MC problem in the absence of noise. Our method not only inherits the advantage of the MC methods, which are based on the MF and thus avoid calculating the SVD, but also can be easily embedded into many other MF-based MC methods, which makes our method more scalable and flexible. More importantly, different from the existing MF-based MC methods, our method imposes smoothness constraints on each of the factor matrices to help select the factor matrices of high quality, directly leading to the smooth MF, which largely improves the MC performance in both the recovery accuracy and the recovery speed. Two kinds of alternating minimization algorithms are also put forward to numerically solve the proposed model, and extensive experiments further confirm the superior performance of the proposed method over many state-of-the-art methods.
- 1 September 2024
- Published by SPIE-Intl Soc Optical Eng in Journal of Electronic Imaging
- Vol. 33 (05) , 053062-18
Exact matrix completion via smooth matrix factorization
Matrix completion (MC) has been successfully applied to many real-world applications, such as recommender systems and image inpainting. Traditional MC methods mainly aim at minimizing the rank function or its surrogate functions. However, the resulting algorithms inevitably calculate the singular value decomposition (SVD) at each iteration, which not only consumes a great deal of time but also greatly limits the potential applications in large-scale data. We propose an MC method that is based on smooth matrix factorization (MF) to deal with the MC problem in the absence of noise. Our method not only inherits the advantage of the MC methods, which are based on the MF and thus avoid calculating the SVD, but also can be easily embedded into many other MF-based MC methods, which makes our method more scalable and flexible. More importantly, different from the existing MF-based MC methods, our method imposes smoothness constraints on each of the factor matrices to help select the factor matrices of high quality, directly leading to the smooth MF, which largely improves the MC performance in both the recovery accuracy and the recovery speed. Two kinds of alternating minimization algorithms are also put forward to numerically solve the proposed model, and extensive experiments further confirm the superior performance of the proposed method over many state-of-the-art methods.
- 9 August 2022
- Published by Elsevier in Signal Processing
- Vol. 201
Performance guarantees of regularized -minimization for robust sparse recovery- 9 August 2022
- Published by Elsevier in Signal Processing
- Vol. 201
Performance guarantees of regularized -minimization for robust sparse recovery- 28 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 22 (7) , 6764-6771
Adsorption Behavior of Pd-Doped PtS₂ Monolayer Upon SF₆ Decomposed Species and the Effect of Applied Electric Field
PtS 2 monolayer with desirable gas sensing property is suggested in this work based on the first-principles theory for detection of three SF 6 decomposed species, in order to evaluate the operation status of the SF 6 insulation devices in power systems, and Pd-doping is proposed to enhance the sensing performance of PtS 2 monolayer. Results indicate that Pd dopant is stably anchored on the $\text{T}_{\rm Pt}$ site of the PtS 2 surface with a ${E}_{\rm b}$ of −2.34 eV. The adsorption performance of the Pd-doped PtS 2 (Pd-PtS 2 ) monolayer upon three SF 6 decomposed species is in order as SO 2 > SOF 2 > SO 2 F 2 , with ${E}_{\rm ad}$ of −1.04, −0.84 and −0.19 eV, respectively. Thus, chemisorption is identified for the SO 2 and SOF 2 systems with remarkable increase of bandgap for the Pd-PtS 2 monolayer by 0.044 and 0.041 eV, respectively, while physisorption is identified for the SO 2 F 2 system with relatively smaller increase of bandgap by 0.029 eV. Moreover, an applied electric field can effectively modulate the charge-transfer for the gas adsorption systems, which manifests the strong potential of Pd-PtS 2 monolayer as a field-effect transistor gas sensor for detection of SO 2 , SOF 2 and SO 2 F 2 with high sensing response. Moreover, I-V curves in Pd-PtS 2 system with and without gas adsorption are also analyzed to verify the above findings. Our theoretical report is meaningful to explore novel sensing materials for application in the field of electrical engineering and to guarantee the good operation of power systems.
- 28 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 22 (7) , 6764-6771
Adsorption Behavior of Pd-Doped PtS₂ Monolayer Upon SF₆ Decomposed Species and the Effect of Applied Electric Field
PtS 2 monolayer with desirable gas sensing property is suggested in this work based on the first-principles theory for detection of three SF 6 decomposed species, in order to evaluate the operation status of the SF 6 insulation devices in power systems, and Pd-doping is proposed to enhance the sensing performance of PtS 2 monolayer. Results indicate that Pd dopant is stably anchored on the $\text{T}_{\rm Pt}$ site of the PtS 2 surface with a ${E}_{\rm b}$ of −2.34 eV. The adsorption performance of the Pd-doped PtS 2 (Pd-PtS 2 ) monolayer upon three SF 6 decomposed species is in order as SO 2 > SOF 2 > SO 2 F 2 , with ${E}_{\rm ad}$ of −1.04, −0.84 and −0.19 eV, respectively. Thus, chemisorption is identified for the SO 2 and SOF 2 systems with remarkable increase of bandgap for the Pd-PtS 2 monolayer by 0.044 and 0.041 eV, respectively, while physisorption is identified for the SO 2 F 2 system with relatively smaller increase of bandgap by 0.029 eV. Moreover, an applied electric field can effectively modulate the charge-transfer for the gas adsorption systems, which manifests the strong potential of Pd-PtS 2 monolayer as a field-effect transistor gas sensor for detection of SO 2 , SOF 2 and SO 2 F 2 with high sensing response. Moreover, I-V curves in Pd-PtS 2 system with and without gas adsorption are also analyzed to verify the above findings. Our theoretical report is meaningful to explore novel sensing materials for application in the field of electrical engineering and to guarantee the good operation of power systems.
- 27 January 2022
- Published by Walter de Gruyter GmbH in Journal of Inverse and Ill-posed Problems
- Vol. 30 (5) , 767-776
Perturbation analysis of 𝐿1‒2 method for robust sparse recovery
One of the crucial challenges in applying the theory of compressed sensing in practice is to cope with the perturbation caused by different types of unknown noise, which may arise from the physical implementation or the human mis-modeling. In this paper, we study the robust recovery from a completely perturbed model via a recently popular method. By using the powerful restricted isometry constant of order for , we first obtain a series of perturbation analysis results for this method, which shows that this method is also able to guarantee a robust recovery for any (nearly) sparse signals when both the ideal observations and the exact measurement matrix are perturbed by the unknown noise. Moreover, one of the established recovery conditions under the noise-free settings is also demonstrated to be much better than the state-of-the-art one. Finally, some simulation experiments are further carried out to verify the effectiveness of this method.- 27 January 2022
- Published by Walter de Gruyter GmbH in Journal of Inverse and Ill-posed Problems
- Vol. 30 (5) , 767-776
Perturbation analysis of 𝐿1‒2 method for robust sparse recovery
One of the crucial challenges in applying the theory of compressed sensing in practice is to cope with the perturbation caused by different types of unknown noise, which may arise from the physical implementation or the human mis-modeling. In this paper, we study the robust recovery from a completely perturbed model via a recently popular method. By using the powerful restricted isometry constant of order for , we first obtain a series of perturbation analysis results for this method, which shows that this method is also able to guarantee a robust recovery for any (nearly) sparse signals when both the ideal observations and the exact measurement matrix are perturbed by the unknown noise. Moreover, one of the established recovery conditions under the noise-free settings is also demonstrated to be much better than the state-of-the-art one. Finally, some simulation experiments are further carried out to verify the effectiveness of this method.- 29 June 2021
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 12 (12) , 3379-3395
Performance guarantees of transformed Schatten-1 regularization for exact low-rank matrix recovery
- 29 June 2021
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 12 (12) , 3379-3395
Performance guarantees of transformed Schatten-1 regularization for exact low-rank matrix recovery
- Open Access
Low-rank matrix recovery via regularized nuclear norm minimization
In this paper, we theoretically investigate the low-rank matrix recovery problem in the context of the unconstrained regularized nuclear norm minimization (RNNM) framework. Our theoretical findings show that, the RNNM method is able to provide a robust recovery of any matrix $X$ (not necessary to be exactly low-rank) from its few noisy measurements $\textbf{b}=\mathcal{A}(X)+\textbf{n}$ with a bounded constraint $\|\textbf{n}\|_{2}\leq\epsilon$, provided that the $tk$-order restricted isometry constant (RIC) of $\mathcal{A}$ satisfies a certain constraint related to $t>0$. Specifically, the obtained recovery condition in the case of $t>4/3$ is found to be same with the sharp condition established previously by Cai and Zhang (2014) to guarantee the exact recovery of any rank-$k$ matrix via the constrained nuclear norm minimization method. More importantly, to the best of our knowledge, we are the first to establish the $tk$-order RIC based coefficient estimate of the robust null space property in the case of $0<t\leq1$.Open AccessLow-rank matrix recovery via regularized nuclear norm minimization
In this paper, we theoretically investigate the low-rank matrix recovery problem in the context of the unconstrained regularized nuclear norm minimization (RNNM) framework. Our theoretical findings show that, the RNNM method is able to provide a robust recovery of any matrix $X$ (not necessary to be exactly low-rank) from its few noisy measurements $\textbf{b}=\mathcal{A}(X)+\textbf{n}$ with a bounded constraint $\|\textbf{n}\|_{2}\leq\epsilon$, provided that the $tk$-order restricted isometry constant (RIC) of $\mathcal{A}$ satisfies a certain constraint related to $t>0$. Specifically, the obtained recovery condition in the case of $t>4/3$ is found to be same with the sharp condition established previously by Cai and Zhang (2014) to guarantee the exact recovery of any rank-$k$ matrix via the constrained nuclear norm minimization method. More importantly, to the best of our knowledge, we are the first to establish the $tk$-order RIC based coefficient estimate of the robust null space property in the case of $0<t\leq1$. - 1 November 2020
- Published by Elsevier in Applied and Computational Harmonic Analysis
- Vol. 49 (3) , 831-862
Group sparse recovery in impulsive noise via alternating direction method of multipliers
- 1 November 2020
- Published by Elsevier in Applied and Computational Harmonic Analysis
- Vol. 49 (3) , 831-862
Group sparse recovery in impulsive noise via alternating direction method of multipliers
- Open Access
- 7 September 2020
- Published by Taylor & Francis in Journal of Applied Statistics
- Vol. 49 (2) , 411-426
Feature selection based on distance correlation: a filter algorithm
Feature selection (FS) is one of the most powerful techniques to cope with the curse of dimensionality. In the study, a new filter approach to feature selection based on distance correlation is presented (DCFS, for short), which keeps the model-free advantage without any pre-specified parameters. Our method consists of two steps: hard step (forward selection) and soft step (backward selection). In the hard step, two types of associations, between univariate feature and the classes and between group feature and the classes, are involved to pick out the most relevant features with respect to the target classes. Due to the strict screening condition in the first step, some of the useful features are likely removed. Therefore, in the soft step, a feature-relationship gain (like feature score) based on the distance correlation is introduced, which is concerned with five kinds of associations. We sort the feature gain values and implement the backward selection procedure until the errors stop declining. The simulation results show that our method becomes more competitive on several datasets compared with some of the representative feature selection methods based on several classification models.
- 7 September 2020
- Published by Taylor & Francis in Journal of Applied Statistics
- Vol. 49 (2) , 411-426
Open AccessFeature selection based on distance correlation: a filter algorithm
Feature selection (FS) is one of the most powerful techniques to cope with the curse of dimensionality. In the study, a new filter approach to feature selection based on distance correlation is presented (DCFS, for short), which keeps the model-free advantage without any pre-specified parameters. Our method consists of two steps: hard step (forward selection) and soft step (backward selection). In the hard step, two types of associations, between univariate feature and the classes and between group feature and the classes, are involved to pick out the most relevant features with respect to the target classes. Due to the strict screening condition in the first step, some of the useful features are likely removed. Therefore, in the soft step, a feature-relationship gain (like feature score) based on the distance correlation is introduced, which is concerned with five kinds of associations. We sort the feature gain values and implement the backward selection procedure until the errors stop declining. The simulation results show that our method becomes more competitive on several datasets compared with some of the representative feature selection methods based on several classification models.