Weipeng Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Weipeng Wu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Farrukh Dekhkonov
1 shared publication- Farrukh Dekhkonov
Department of Mathematics, Namangan State University, Namangan, Uzbekistan
1 shared publication - He Ma
School of Computer and Information Engineering, Heilongjiang University of Science and Technology, Harbin 150022, China
1 shared publication - Jiangbo Han
School of Mathematical Sciences, Inner Mongolia University , 010021 Hohhot , People’s Republic of China
1 shared publication - Juan Zhang
School of Software, Harbin Institute of Information Technology, No. 9, University Town, Binxi Technology Development Zone, Harbin 150431, China
1 shared publication Malika Nematjonova
1 shared publication
Latest Publications
- Open Access
- 20 August 2026
- Published by MDPI AG in Electronics
- Vol. 15 (16) , 3725
Masking-Guided Structure and Texture Decoupling for Lightweight Blind Screen Content Image Quality Assessment
Screen content images (SCIs) exhibit complex structural heterogeneity, rendering traditional statistics-based natural scene image quality assessment (NR-IQA) metrics ineffective. Although deep learning models achieve high prediction accuracy, their prohibitive computational demands preclude deployment in latency-sensitive industrial scenarios. While existing handcrafted lightweight SCI-IQA metrics reduce computational overhead, most rely on unsegmented global feature pooling or holistic edge statistics (e.g., edge histograms or Fisher vector coding), thereby diluting locally critical text-edge distortions in vast homogeneous backgrounds. To address this limitation, we propose an ultra-lightweight, deep-learning-free NR-IQA framework centered on human visual masking. Unlike existing lightweight methods, our approach explicitly employs dual-scale Canny edge operators to partition SCIs into edge-sensitive and flat background regions. Guided by this visual prior, structural degradations and micro-compression textures are extracted region-wise using Sobel gradients and uniform local binary patterns (LBPs) and aggregated with global Commission Internationale de I’Eclairage L*a*b*(CIELAB) color statistics into a compact 60-dimensional descriptor. A grid-search-optimized Support Vector Regression (SVR) maps these features to subjective quality scores. Extensive cross-validation on the SIQAD and SCID datasets demonstrates that our metric outperforms existing handcrafted lightweight SCI metrics and traditional NSS models, while achieving accuracy competitive with representative full-reference metrics. Consuming only 79.3 ms per image on a standard CPU, it offers a practical accuracy–efficiency trade-off for resource-constrained periodic quality monitoring.
- 20 August 2026
- Published by MDPI AG in Electronics
- Vol. 15 (16) , 3725
Open AccessMasking-Guided Structure and Texture Decoupling for Lightweight Blind Screen Content Image Quality Assessment
Screen content images (SCIs) exhibit complex structural heterogeneity, rendering traditional statistics-based natural scene image quality assessment (NR-IQA) metrics ineffective. Although deep learning models achieve high prediction accuracy, their prohibitive computational demands preclude deployment in latency-sensitive industrial scenarios. While existing handcrafted lightweight SCI-IQA metrics reduce computational overhead, most rely on unsegmented global feature pooling or holistic edge statistics (e.g., edge histograms or Fisher vector coding), thereby diluting locally critical text-edge distortions in vast homogeneous backgrounds. To address this limitation, we propose an ultra-lightweight, deep-learning-free NR-IQA framework centered on human visual masking. Unlike existing lightweight methods, our approach explicitly employs dual-scale Canny edge operators to partition SCIs into edge-sensitive and flat background regions. Guided by this visual prior, structural degradations and micro-compression textures are extracted region-wise using Sobel gradients and uniform local binary patterns (LBPs) and aggregated with global Commission Internationale de I’Eclairage L*a*b*(CIELAB) color statistics into a compact 60-dimensional descriptor. A grid-search-optimized Support Vector Regression (SVR) maps these features to subjective quality scores. Extensive cross-validation on the SIQAD and SCID datasets demonstrates that our metric outperforms existing handcrafted lightweight SCI metrics and traditional NSS models, while achieving accuracy competitive with representative full-reference metrics. Consuming only 79.3 ms per image on a standard CPU, it offers a practical accuracy–efficiency trade-off for resource-constrained periodic quality monitoring.
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Communications in Analysis and Mechanics
- Vol. 18 (3) , 732-748
On the time-optimal control problem for a pseudo-parabolic equation with variable thermal conductivity in a bounded domain
This paper studied a time-optimal control problem for a pseudo-parabolic equation in a bounded multidimensional domain. The equation involved a uniformly elliptic operator and a small parameter characterizing memory effects. The control acted through a distributed source term, where the spatial part was fixed and the time-dependent part was a bounded measurable function. The goal was to steer the system from a zero initial state to a prescribed spatially averaged temperature in minimal time and maintain it thereafter. Using spectral decomposition, we derived an explicit solution representation and reduced the control problem to a Volterra integral equation. Under natural positivity assumptions on the Fourier coefficients, we established that the kernel was continuous, positive, and monotone. These properties yielded an explicit upper bound for the minimal time in terms of the first eigenvalue, the relaxation parameter, and the system constraints. A comparison with the parabolic case showed that the pseudo-parabolic term led to longer minimal times, reflecting the memory effects inherent in such models.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Communications in Analysis and Mechanics
- Vol. 18 (3) , 732-748
Open AccessOn the time-optimal control problem for a pseudo-parabolic equation with variable thermal conductivity in a bounded domain
This paper studied a time-optimal control problem for a pseudo-parabolic equation in a bounded multidimensional domain. The equation involved a uniformly elliptic operator and a small parameter characterizing memory effects. The control acted through a distributed source term, where the spatial part was fixed and the time-dependent part was a bounded measurable function. The goal was to steer the system from a zero initial state to a prescribed spatially averaged temperature in minimal time and maintain it thereafter. Using spectral decomposition, we derived an explicit solution representation and reduced the control problem to a Volterra integral equation. Under natural positivity assumptions on the Fourier coefficients, we established that the kernel was continuous, positive, and monotone. These properties yielded an explicit upper bound for the minimal time in terms of the first eigenvalue, the relaxation parameter, and the system constraints. A comparison with the parabolic case showed that the pseudo-parabolic term led to longer minimal times, reflecting the memory effects inherent in such models.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Discrete and Continuous Dynamical Systems - S
- Vol. 29, 53-70
Boundary control problem for a heat equation in a multidimensional parallelepiped
In this paper, we consider a boundary control problem associated with the heat equation in a bounded $ n $-dimensional domain. The value of the solution is prescribed on a part of the boundary, and the objective is to determine a control function that ensures a prescribed average temperature is attained within the domain. The initial-boundary problem is solved by the Fourier method, and the control problem under consideration is that of the Volterra integral equation. The existence of admissible control is proved by the Laplace transform method.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Discrete and Continuous Dynamical Systems - S
- Vol. 29, 53-70
Boundary control problem for a heat equation in a multidimensional parallelepiped
In this paper, we consider a boundary control problem associated with the heat equation in a bounded $ n $-dimensional domain. The value of the solution is prescribed on a part of the boundary, and the objective is to determine a control function that ensures a prescribed average temperature is attained within the domain. The initial-boundary problem is solved by the Fourier method, and the control problem under consideration is that of the Volterra integral equation. The existence of admissible control is proved by the Laplace transform method.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Discrete and Continuous Dynamical Systems - S
- Vol. 22, 293-326
Ordered sample weighting: Improving neural network training for complex image classification
The growing abundance of labeled image data has propelled significant advancements in neural networks, particularly in image classification tasks, where automatic feature extraction plays a crucial role. However, the inherent imbalance in data distribution and the varying complexity of samples in real-world applications hinder consistent classification performance across different classes. This paper addresses the challenges by proposing a classification loss-weighted neural network training and optimization method. The method begins by selecting a batch of samples and ranking them based on classification loss values. Hard-to-classify samples with larger loss values are prioritized, reducing the repetitive training of easy samples. The proposed Top-q Sample Weighting (TSW) strategy estimates sample weights by maximizing the reduction in classification loss, focusing training on the most informative samples. The weighted sample subset is used to update the network parameters, improving both convergence speed and classification performance. Comparative experiments on five datasets and two network architectures validate the effectiveness of the proposed method, demonstrating significant improvements in neural network generalization and classification accuracy.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Discrete and Continuous Dynamical Systems - S
- Vol. 22, 293-326
Ordered sample weighting: Improving neural network training for complex image classification
The growing abundance of labeled image data has propelled significant advancements in neural networks, particularly in image classification tasks, where automatic feature extraction plays a crucial role. However, the inherent imbalance in data distribution and the varying complexity of samples in real-world applications hinder consistent classification performance across different classes. This paper addresses the challenges by proposing a classification loss-weighted neural network training and optimization method. The method begins by selecting a batch of samples and ranking them based on classification loss values. Hard-to-classify samples with larger loss values are prioritized, reducing the repetitive training of easy samples. The proposed Top-q Sample Weighting (TSW) strategy estimates sample weights by maximizing the reduction in classification loss, focusing training on the most informative samples. The weighted sample subset is used to update the network parameters, improving both convergence speed and classification performance. Comparative experiments on five datasets and two network architectures validate the effectiveness of the proposed method, demonstrating significant improvements in neural network generalization and classification accuracy.
- Open Access
- 1 January 2025
- Published by Walter de Gruyter GmbH in Advances in Nonlinear Analysis
- Vol. 14 (1)
Continuous dependence on initial data for damped fourth-order wave equation with strain term
This article investigates the fourth-order wave equation with general nonlinear strain term as well as both strong and weak damping terms. By deriving and utilizing energy decay estimate independent of the initial data in the stable manifold, we derive results regarding the continuous dependence of the global solution on initial data for the model equation with such a structure. These conclusions reflect the dissipative nature of the model.
- 1 January 2025
- Published by Walter de Gruyter GmbH in Advances in Nonlinear Analysis
- Vol. 14 (1)
Open AccessContinuous dependence on initial data for damped fourth-order wave equation with strain term
This article investigates the fourth-order wave equation with general nonlinear strain term as well as both strong and weak damping terms. By deriving and utilizing energy decay estimate independent of the initial data in the stable manifold, we derive results regarding the continuous dependence of the global solution on initial data for the model equation with such a structure. These conclusions reflect the dissipative nature of the model.
- Open Access
- 1 January 2024
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 32 (6) , 3952-3972
A deep clustering framework integrating pairwise constraints and a VMF mixture model
We presented a novel deep generative clustering model called Variational Deep Embedding based on Pairwise constraints and the Von Mises-Fisher mixture model (VDEPV). VDEPV consists of fully connected neural networks capable of learning latent representations from raw data and accurately predicting cluster assignments. Under the assumption of a genuinely non-informative prior, VDEPV adopted a von Mises-Fisher mixture model to depict the hyperspherical interpretation of the data. We defined and established pairwise constraints by employing a random sample mining strategy and applying data augmentation techniques. These constraints enhanced the compactness of intra-cluster samples in the spherical embedding space while improving inter-cluster samples' separability. By minimizing Kullback-Leibler divergence, we formulated a clustering loss function based on pairwise constraints, which regularized the joint probability distribution of latent variables and cluster labels. Comparative experiments with other deep clustering methods demonstrated the excellent performance of VDEPV.
- 1 January 2024
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 32 (6) , 3952-3972
Open AccessA deep clustering framework integrating pairwise constraints and a VMF mixture model
We presented a novel deep generative clustering model called Variational Deep Embedding based on Pairwise constraints and the Von Mises-Fisher mixture model (VDEPV). VDEPV consists of fully connected neural networks capable of learning latent representations from raw data and accurately predicting cluster assignments. Under the assumption of a genuinely non-informative prior, VDEPV adopted a von Mises-Fisher mixture model to depict the hyperspherical interpretation of the data. We defined and established pairwise constraints by employing a random sample mining strategy and applying data augmentation techniques. These constraints enhanced the compactness of intra-cluster samples in the spherical embedding space while improving inter-cluster samples' separability. By minimizing Kullback-Leibler divergence, we formulated a clustering loss function based on pairwise constraints, which regularized the joint probability distribution of latent variables and cluster labels. Comparative experiments with other deep clustering methods demonstrated the excellent performance of VDEPV.