Zheng Kou
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zheng Kou has a total of 82 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoli Qiang
Guangzhou University
13 shared publications - Qura Tul Ain
Department of Mathematics, Guizhou University, 550025 Guiyang, P. R. China
5 shared publications - Maryam Akhoundi
Clinical Research Development Unit of Rouhani Hospital, Babol University of Medical Sciences, Babol, Iran
4 shared publications - Saeed Kosari
Institute of Computing Science and Technology, Guangzhou University, Guangzhou 510006, P. R. China
4 shared publications - Xiaolong Shi
Institute of Computing Science and Technology, Guangzhou University, China
4 shared publications - Fang Wang
Guangzhou University
3 shared publications
Latest Publications
- Open Access
- 28 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (5) , 4002-4012
Identification of Pandemic Risk for Avian Influenza Virus With Graph Cross Attention Networks
Influenza A virus poses a significant threat to global health due to its high mutability and potential for pandemics. Accurate prediction of influenza virus pandemic risk can mitigate their threat to public health systems and societal stability. Current sequence-based deep learning methods predict the risk of viral pandemics by extracting features from the viral genome sequence. However, these methods ignore the gene flow between different virus strains through genetic reassortment, which is essential for understanding the dynamics of viral evolution. To leverage the gene flow of virus strains to predict the pandemic, we propose Graph Cross Attention Networks for influenza pandemic identification (GCAN-Flu). GCAN-Flu begins with constructing an influenza virus gene flow network based on the genetic reassortment relationship between viruses, and then encodes the gene flow information between strains as a latent modal representation. We further use Graph Cross Attention (GCA) layers to integrate the viral genome sequence with the gene flow information, and predict pandemic risks based on the multimodal representations. Experimental results demonstrate that GCAN-Flu outperforms current methods, providing a powerful tool for pandemic forecasting. All source code and datasets are publicly available at https://github.com/kouzheng/GCAN.
- 28 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (5) , 4002-4012
Open AccessIdentification of Pandemic Risk for Avian Influenza Virus With Graph Cross Attention Networks
Influenza A virus poses a significant threat to global health due to its high mutability and potential for pandemics. Accurate prediction of influenza virus pandemic risk can mitigate their threat to public health systems and societal stability. Current sequence-based deep learning methods predict the risk of viral pandemics by extracting features from the viral genome sequence. However, these methods ignore the gene flow between different virus strains through genetic reassortment, which is essential for understanding the dynamics of viral evolution. To leverage the gene flow of virus strains to predict the pandemic, we propose Graph Cross Attention Networks for influenza pandemic identification (GCAN-Flu). GCAN-Flu begins with constructing an influenza virus gene flow network based on the genetic reassortment relationship between viruses, and then encodes the gene flow information between strains as a latent modal representation. We further use Graph Cross Attention (GCA) layers to integrate the viral genome sequence with the gene flow information, and predict pandemic risks based on the multimodal representations. Experimental results demonstrate that GCAN-Flu outperforms current methods, providing a powerful tool for pandemic forecasting. All source code and datasets are publicly available at https://github.com/kouzheng/GCAN.
- 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 56 (1) , 523-536
Less Is More: A Small-Scale Learning Particle Swarm Optimization for Large-Scale Optimization
Large-scale optimization problem (LSOP) is an essential research topic in the field of evolutionary computation community. Many large-scale optimization algorithms often maintain a large population for diversity enhancement. However, updating such a large population consumes a significant number of fitness evaluations (FEs), which may lead to the insufficient evolution of the population. In light of this, this article proposes a small-scale learning particle swarm optimization (SSLPSO) for solving LSOPs. In the small-scale learning mechanism, only up to two representative individuals are updated in every generation to effectively save FEs and prolong the evolutionary generations, so as to refine the solution accuracy. Specifically, we first design a representative individual selection (RIS) strategy to select the convergence representative individual and the diversity representative individual for updating. Then, we develop a representative individual learning (RIL) strategy, which includes a convergence learning method and a diversity learning method for the convergence representative individual and the diversity representative individual, respectively. Meanwhile, we further propose an adaptive strategy adjustment (ASA) method based on evolutionary state assessment to determine whether the representative individuals should be updated, further achieving the adaptive adjustment of the evolutionary behavior in the population. Experimental results on the commonly used large-scale test suites, IEEE CEC2010 and IEEE CEC2013, show that the performance of SSLPSO is significantly better than, or at least comparable to other state-of-the-art large-scale optimization algorithms, including the winners of large-scale competitions. Finally, the application of SSLPSO to a large-scale constrained water distribution network optimization problem further demonstrates its real-world applicability.
- 17 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 56 (1) , 523-536
Less Is More: A Small-Scale Learning Particle Swarm Optimization for Large-Scale Optimization
Large-scale optimization problem (LSOP) is an essential research topic in the field of evolutionary computation community. Many large-scale optimization algorithms often maintain a large population for diversity enhancement. However, updating such a large population consumes a significant number of fitness evaluations (FEs), which may lead to the insufficient evolution of the population. In light of this, this article proposes a small-scale learning particle swarm optimization (SSLPSO) for solving LSOPs. In the small-scale learning mechanism, only up to two representative individuals are updated in every generation to effectively save FEs and prolong the evolutionary generations, so as to refine the solution accuracy. Specifically, we first design a representative individual selection (RIS) strategy to select the convergence representative individual and the diversity representative individual for updating. Then, we develop a representative individual learning (RIL) strategy, which includes a convergence learning method and a diversity learning method for the convergence representative individual and the diversity representative individual, respectively. Meanwhile, we further propose an adaptive strategy adjustment (ASA) method based on evolutionary state assessment to determine whether the representative individuals should be updated, further achieving the adaptive adjustment of the evolutionary behavior in the population. Experimental results on the commonly used large-scale test suites, IEEE CEC2010 and IEEE CEC2013, show that the performance of SSLPSO is significantly better than, or at least comparable to other state-of-the-art large-scale optimization algorithms, including the winners of large-scale competitions. Finally, the application of SSLPSO to a large-scale constrained water distribution network optimization problem further demonstrates its real-world applicability.
- Open Access
- 17 June 2025
- Published by MDPI AG in Fractal and Fractional
- Vol. 9 (6) , 387
Spatiotemporal Analysis of Disease Spread Using a Soliton-Based SIR Framework for Nomadic Populations
This study enhances the classical deterministic SIR model by incorporating soliton-like dynamics and gradient-induced diffusion, effectively capturing the complex spatiotemporal patterns of disease transmission within nomadic populations. The proposed model incorporates an advection–diffusion mechanism that modulates the spatial gradients in infection dynamics, transitioning from highly localized infection peaks to distributed infection fronts. We discussed the role of diffusion coefficients in shaping the spatial distribution of susceptible, infected, and recovered populations, as well as the impact of gradient-induced advection in mitigating epidemic intensity. Numerical simulations demonstrate the effects of varying key parameters such as transmission rates, recovery rates, and advection–diffusion coefficients on the epidemic’s progression. The soliton-like dynamics ensure the stability of infection waves over time, specifying targeted intervention strategies such as localized quarantines and vaccination campaigns. This model underscores the critical importance of spatial heterogeneity and mobility patterns in managing infectious diseases. The applicability of the model has been tested using the AIDS data from the last 25 years.
- 17 June 2025
- Published by MDPI AG in Fractal and Fractional
- Vol. 9 (6) , 387
Open AccessSpatiotemporal Analysis of Disease Spread Using a Soliton-Based SIR Framework for Nomadic Populations
This study enhances the classical deterministic SIR model by incorporating soliton-like dynamics and gradient-induced diffusion, effectively capturing the complex spatiotemporal patterns of disease transmission within nomadic populations. The proposed model incorporates an advection–diffusion mechanism that modulates the spatial gradients in infection dynamics, transitioning from highly localized infection peaks to distributed infection fronts. We discussed the role of diffusion coefficients in shaping the spatial distribution of susceptible, infected, and recovered populations, as well as the impact of gradient-induced advection in mitigating epidemic intensity. Numerical simulations demonstrate the effects of varying key parameters such as transmission rates, recovery rates, and advection–diffusion coefficients on the epidemic’s progression. The soliton-like dynamics ensure the stability of infection waves over time, specifying targeted intervention strategies such as localized quarantines and vaccination campaigns. This model underscores the critical importance of spatial heterogeneity and mobility patterns in managing infectious diseases. The applicability of the model has been tested using the AIDS data from the last 25 years.
- Top Cited PapersOpen Access
- 8 February 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-13
GTAT: empowering graph neural networks with cross attention
Graph Neural Networks (GNNs) serve as a powerful framework for representation learning on graph-structured data, capturing the information of nodes by recursively aggregating and transforming the neighboring nodes’ representations. Topology in graph plays an important role in learning graph representations and impacts the performance of GNNs. However, current methods fail to adequately integrate topological information into graph representation learning. To better leverage topological information and enhance representation capabilities, we propose the Graph Topology Attention Networks (GTAT). Specifically, GTAT first extracts topology features from the graph’s structure and encodes them into topology representations. Then, the representations of node and topology are fed into cross attention GNN layers for interaction. This integration allows the model to dynamically adjust the influence of node features and topological information, thus improving the expressiveness of nodes. Experimental results on various graph benchmark datasets demonstrate GTAT outperforms recent state-of-the-art methods. Further analysis reveals GTAT’s capability to mitigate the over-smoothing issue, and its increased robustness against noisy data.
- 8 February 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-13
Top Cited PapersOpen AccessGTAT: empowering graph neural networks with cross attention
Graph Neural Networks (GNNs) serve as a powerful framework for representation learning on graph-structured data, capturing the information of nodes by recursively aggregating and transforming the neighboring nodes’ representations. Topology in graph plays an important role in learning graph representations and impacts the performance of GNNs. However, current methods fail to adequately integrate topological information into graph representation learning. To better leverage topological information and enhance representation capabilities, we propose the Graph Topology Attention Networks (GTAT). Specifically, GTAT first extracts topology features from the graph’s structure and encodes them into topology representations. Then, the representations of node and topology are fed into cross attention GNN layers for interaction. This integration allows the model to dynamically adjust the influence of node features and topological information, thus improving the expressiveness of nodes. Experimental results on various graph benchmark datasets demonstrate GTAT outperforms recent state-of-the-art methods. Further analysis reveals GTAT’s capability to mitigate the over-smoothing issue, and its increased robustness against noisy data.
- Open Access
- 19 December 2024
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (1)
Instruction-responsive programmable assemblies with DNA origami block pieces
DNA nanotechnology has created a wide variety of nanostructures that provide a reliable platform for nanofabrication and DNA computing. However, constructing programmable finite arrays that allow for easy pre-functionalization remains challenge. We aim to create more standardized and controllable DNA origami components, which could be assembled into finite-scale and more diverse superstructures driven by instruction sets. In this work, we designed and implemented DNA origami building block pieces (DOBPs) with eight mutually independent programmable edges and formulated DNA instructions that tailored such components. This system enables DOBPs to be assembled into one or more specific 2D arrays according to the instruction sets. Theoretically, a two-unit system can generate up to 48 distinct DNA arrays. Importantly, experiments results demonstrated that DOBPs are capable of both deterministic and nondeterministic assemblies. Moreover, after examining the effects of different connection strategies and instruction implementations on the yield of the target structures, we assembled more complex 2D arrays, including limited self-assembly arrays such as ‘square frames’, ‘windmills’ and ‘multiples of 3’ long strips. We also demonstrated examples of Boolean logic gates ‘AND’ and ‘XOR’ computations based on these assembly arrays. The assembly system provides a model nano-structure for the research on controllable finite self-assembly and offers a more integrated approach for the storage and processing of molecular information.
- 19 December 2024
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (1)
Open AccessInstruction-responsive programmable assemblies with DNA origami block pieces
DNA nanotechnology has created a wide variety of nanostructures that provide a reliable platform for nanofabrication and DNA computing. However, constructing programmable finite arrays that allow for easy pre-functionalization remains challenge. We aim to create more standardized and controllable DNA origami components, which could be assembled into finite-scale and more diverse superstructures driven by instruction sets. In this work, we designed and implemented DNA origami building block pieces (DOBPs) with eight mutually independent programmable edges and formulated DNA instructions that tailored such components. This system enables DOBPs to be assembled into one or more specific 2D arrays according to the instruction sets. Theoretically, a two-unit system can generate up to 48 distinct DNA arrays. Importantly, experiments results demonstrated that DOBPs are capable of both deterministic and nondeterministic assemblies. Moreover, after examining the effects of different connection strategies and instruction implementations on the yield of the target structures, we assembled more complex 2D arrays, including limited self-assembly arrays such as ‘square frames’, ‘windmills’ and ‘multiples of 3’ long strips. We also demonstrated examples of Boolean logic gates ‘AND’ and ‘XOR’ computations based on these assembly arrays. The assembly system provides a model nano-structure for the research on controllable finite self-assembly and offers a more integrated approach for the storage and processing of molecular information.
- Open Access
- 26 November 2024
- Published by Frontiers Media SA in Frontiers in Physics
- Vol. 12, 1500423
Stochastic modeling of plant-insect interaction dynamics with MEMS-based monitoring and noise effects
The dynamics of plant-insect interactions play a crucial role in the ecosystem, influenced by complex molecular signaling pathways. This study extends existing deterministic models of plant-insect systems by incorporating stochastic elements and molecular interactions, particularly focusing on the roles of Botrytis Induced Kinase-1 (BIK1) and Phyto Alexin Deficient-4 (PAD4) proteins. The model evaluates the effects of constant inhibition, pulsed inhibition, and adaptive feedback control on plant biomass , insect herbivore density , PAD4 levels , and BIK1 levels . Additionally, we examine the impact of different noise types, including deterministic, Gaussian, and Lévy noise, on system variability and stability. Results indicate that our stochastic model is superior as it shows a significant reduction in BIK1 levels, particularly under higher noise intensities, which enhances PAD4 activity and improves plant defense mechanisms. Moreover, moderate noise intensity provides an optimal balance, sustaining PAD4 levels while effectively controlling insect herbivore populations. We also integrate MEMS-based feedback mechanisms, which dynamically adjust plant biomass and molecular signaling, further stabilizing the system’s response to environmental variability.- 26 November 2024
- Published by Frontiers Media SA in Frontiers in Physics
- Vol. 12, 1500423
Open AccessStochastic modeling of plant-insect interaction dynamics with MEMS-based monitoring and noise effects
The dynamics of plant-insect interactions play a crucial role in the ecosystem, influenced by complex molecular signaling pathways. This study extends existing deterministic models of plant-insect systems by incorporating stochastic elements and molecular interactions, particularly focusing on the roles of Botrytis Induced Kinase-1 (BIK1) and Phyto Alexin Deficient-4 (PAD4) proteins. The model evaluates the effects of constant inhibition, pulsed inhibition, and adaptive feedback control on plant biomass , insect herbivore density , PAD4 levels , and BIK1 levels . Additionally, we examine the impact of different noise types, including deterministic, Gaussian, and Lévy noise, on system variability and stability. Results indicate that our stochastic model is superior as it shows a significant reduction in BIK1 levels, particularly under higher noise intensities, which enhances PAD4 activity and improves plant defense mechanisms. Moreover, moderate noise intensity provides an optimal balance, sustaining PAD4 levels while effectively controlling insect herbivore populations. We also integrate MEMS-based feedback mechanisms, which dynamically adjust plant biomass and molecular signaling, further stabilizing the system’s response to environmental variability. - Open Access
- 1 November 2024
- Published by MDPI AG in Biomimetics
- Vol. 9 (11) , 664
Multitask Level-Based Learning Swarm Optimizer
Evolutionary multitasking optimization (EMTO) is currently one of the hottest research topics that aims to utilize the correlation between tasks to optimize them simultaneously. Although many evolutionary multitask algorithms (EMTAs) based on traditional differential evolution (DE) and the genetic algorithm (GA) have been proposed, there are relatively few EMTAs based on particle swarm optimization (PSO). Compared with DE and GA, PSO has a faster convergence speed, especially during the later state of the evolutionary process. Therefore, this paper proposes a multitask level-based learning swarm optimizer (MTLLSO). In MTLLSO, multiple populations are maintained and each population corresponds to the optimization of one task separately using LLSO, leveraging high-level individuals with better fitness to guide the evolution of low-level individuals with worse fitness. When information transfer occurs, high-level individuals from a source population are used to guide the evolution of low-level individuals in the target population to facilitate the effectiveness of knowledge transfer. In this way, MTLLSO can obtain the satisfying balance between self-evolution and knowledge transfer. We have illustrated the effectiveness of MTLLSO on the CEC2017 benchmark, where MTLLSO significantly outperformed other compared algorithms in most problems.
- 1 November 2024
- Published by MDPI AG in Biomimetics
- Vol. 9 (11) , 664
Open AccessMultitask Level-Based Learning Swarm Optimizer
Evolutionary multitasking optimization (EMTO) is currently one of the hottest research topics that aims to utilize the correlation between tasks to optimize them simultaneously. Although many evolutionary multitask algorithms (EMTAs) based on traditional differential evolution (DE) and the genetic algorithm (GA) have been proposed, there are relatively few EMTAs based on particle swarm optimization (PSO). Compared with DE and GA, PSO has a faster convergence speed, especially during the later state of the evolutionary process. Therefore, this paper proposes a multitask level-based learning swarm optimizer (MTLLSO). In MTLLSO, multiple populations are maintained and each population corresponds to the optimization of one task separately using LLSO, leveraging high-level individuals with better fitness to guide the evolution of low-level individuals with worse fitness. When information transfer occurs, high-level individuals from a source population are used to guide the evolution of low-level individuals in the target population to facilitate the effectiveness of knowledge transfer. In this way, MTLLSO can obtain the satisfying balance between self-evolution and knowledge transfer. We have illustrated the effectiveness of MTLLSO on the CEC2017 benchmark, where MTLLSO significantly outperformed other compared algorithms in most problems.
- Open Access
- 8 October 2024
- Published by MDPI AG in Biomimetics
- Vol. 9 (10) , 604
Adaptive Bi-Operator Evolution for Multitasking Optimization Problems
The field of evolutionary multitasking optimization (EMTO) has been a highly anticipated research topic in recent years. EMTO aims to utilize evolutionary algorithms to concurrently solve complex problems involving multiple tasks. Despite considerable advancements in this field, numerous evolutionary multitasking algorithms continue to use a single evolutionary search operator (ESO) throughout the evolution process. This strategy struggles to completely adapt to different tasks, consequently hindering the algorithm’s performance. To overcome this challenge, this paper proposes multitasking evolutionary algorithms via an adaptive bi-operator strategy (BOMTEA). BOMTEA adopts a bi-operator strategy and adaptively controls the selection probability of each ESO according to its performance, which can determine the most suitable ESO for various tasks. In an experiment, BOMTEA showed outstanding results on two well-known multitasking benchmark tests, CEC17 and CEC22, and significantly outperformed other comparative algorithms.
- 8 October 2024
- Published by MDPI AG in Biomimetics
- Vol. 9 (10) , 604
Open AccessAdaptive Bi-Operator Evolution for Multitasking Optimization Problems
The field of evolutionary multitasking optimization (EMTO) has been a highly anticipated research topic in recent years. EMTO aims to utilize evolutionary algorithms to concurrently solve complex problems involving multiple tasks. Despite considerable advancements in this field, numerous evolutionary multitasking algorithms continue to use a single evolutionary search operator (ESO) throughout the evolution process. This strategy struggles to completely adapt to different tasks, consequently hindering the algorithm’s performance. To overcome this challenge, this paper proposes multitasking evolutionary algorithms via an adaptive bi-operator strategy (BOMTEA). BOMTEA adopts a bi-operator strategy and adaptively controls the selection probability of each ESO according to its performance, which can determine the most suitable ESO for various tasks. In an experiment, BOMTEA showed outstanding results on two well-known multitasking benchmark tests, CEC17 and CEC22, and significantly outperformed other comparative algorithms.
- Open Access
- 3 October 2024
- Published by MDPI AG in Fractal and Fractional
- Vol. 8 (10) , 586
Extinction Dynamics and Equilibrium Patterns in Stochastic Epidemic Model for Norovirus: Role of Temporal Immunity and Generalized Incidence Rates
Norovirus is a leading global cause of viral gastroenteritis, significantly affecting mortality, morbidity, and healthcare costs. This paper develops and analyzes a stochastic epidemic model for norovirus dynamics, incorporating temporal immunity and a generalized incidence rate. The model is proven to have a unique positive global solution, with extinction conditions explored. Using Khasminskii’s method, the model’s ergodicity and equilibrium distribution are investigated, demonstrating a unique ergodic stationary distribution when . Extinction occurs when . Computer simulations confirm that noise level significantly influences epidemic spread.- 3 October 2024
- Published by MDPI AG in Fractal and Fractional
- Vol. 8 (10) , 586
Open AccessExtinction Dynamics and Equilibrium Patterns in Stochastic Epidemic Model for Norovirus: Role of Temporal Immunity and Generalized Incidence Rates
Norovirus is a leading global cause of viral gastroenteritis, significantly affecting mortality, morbidity, and healthcare costs. This paper develops and analyzes a stochastic epidemic model for norovirus dynamics, incorporating temporal immunity and a generalized incidence rate. The model is proven to have a unique positive global solution, with extinction conditions explored. Using Khasminskii’s method, the model’s ergodicity and equilibrium distribution are investigated, demonstrating a unique ergodic stationary distribution when . Extinction occurs when . Computer simulations confirm that noise level significantly influences epidemic spread.