Graph Neural Networks(GNNs),as a deep learning framework specifically designed for graph-structured data,have achieved deep representation learning of graph data through message passing mechanisms and have become a co...Graph Neural Networks(GNNs),as a deep learning framework specifically designed for graph-structured data,have achieved deep representation learning of graph data through message passing mechanisms and have become a core technology in the field of graph analysis.However,current reviews on GNN models are mainly focused on smaller domains,and there is a lack of systematic reviews on the classification and applications of GNN models.This review systematically synthesizes the three canonical branches of GNN,Graph Convolutional Network(GCN),Graph Attention Network(GAT),and Graph Sampling Aggregation Network(GraphSAGE),and analyzes their integration pathways from both structural and feature perspectives.Drawing on representative studies,we identify three major integration patterns:cascaded fusion,where heterogeneous modules such as Convolutional Neural Network(CNN),Long Short-Term Memory(LSTM),and GraphSAGE are sequentially combined for hierarchical feature learning;parallel fusion,where multi-branch architectures jointly encode complementary graph features;and feature-level fusion,which employs concatenation,weighted summation,or attention-based gating to adaptively merge multi-source embeddings.Through these patterns,integrated GNNs achieve enhanced expressiveness,robustness,and scalability across domains including transportation,biomedicine,and cybersecurity.展开更多
Robust teleoperation in image-guided interventions faces critical challenges from latency,deformation,and the quasi-periodic nature of physiological motion.This paper presents a fully integrated,latency-aware visual s...Robust teleoperation in image-guided interventions faces critical challenges from latency,deformation,and the quasi-periodic nature of physiological motion.This paper presents a fully integrated,latency-aware visual servoing system leveraging stereo vision,hand–eye calibration,and learning-based prediction for motion-compensated teleoperation.The system combines a calibrated binocular camera setup,dual robotic arms,and a predictive control loop incorporating Long Short-Term Memory(LSTM)and Temporal Convolutional Network(TCN)models.Through experiments using both in vivo and phantom datasets,we quantitatively assess the prediction accuracy and motion-compensation performance of both models.Results show that TCNs deliver more stable and precise tracking,especially on regular trajectories,while LSTMs exhibit robustness under quasi-periodic dynamics.By matching prediction horizons to system latency,the approach significantly reduces peak and steady-state tracking errors,demonstrating practical feasibility for deploying prediction-augmented servoing in teleoperated surgical.展开更多
Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relati...Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relationships among nodes.This paper proposes a novel graph coupling convolutional model that introduces an adaptive weighting mechanism to assign distinct importance to neighboring nodes based on their similarity to the central node.Unlike traditional methods,the proposed coupling strategy enhances the interpretability of node interactions while maintaining competitive classification performance.The model operates in the spatial domain,utilizing adjacency list structures for efficient convolution and addressing the limitations of weight sharing through a coupling-based similarity computation.Extensive experiments are conducted on five graph-structured datasets,including Cora,Citeseer,PubMed,Reddit,and BlogCatalog,as well as a custom topology dataset constructed from the Open University Learning Analytics Dataset(OULAD)educational platform.Results demonstrate that the proposed model achieves good classification accuracy,while significantly reducing training time through direct second-order neighbor fusion and data preprocessing.Moreover,analysis of neighborhood order reveals that considering third-order neighbors offers limited accuracy gains but introduces considerable computational overhead,confirming the efficiency of first-and second-order convolution in practical applications.Overall,the proposed graph coupling model offers a lightweight,interpretable,and effective framework for multi-label node classification in complex networks.展开更多
Background As the population in China rapidly ages,the prevalence of mild cognitive impairment(MCI)is increasing considerably.However,the causes of MCI vary.The continued lack of understanding of the various subtypes ...Background As the population in China rapidly ages,the prevalence of mild cognitive impairment(MCI)is increasing considerably.However,the causes of MCI vary.The continued lack of understanding of the various subtypes of MCI impedes the implementation of effective measures to reduce the risk of advancing to more severe cognitive diseases.Aims To estimate the prevalence and incidence rates of two MCI subtypes—amnestic MCI(aMCI)and vascular cognitive impairment without dementia(VCIND)—and to determine modifiable factors for them among older individuals in a multiregional Chinese cohort.Method This 1-year longitudinal study surveyed a random sample of participants aged≥60 years from a large,community-dwelling cohort in China.Baseline lifestyle data were self-reported,while vascular and comorbid conditions were obtained from medical records and physical examinations.In total,3514 and 2051 individuals completed the baseline and 1-year follow-up assessments,respectively.Logistic and linear regression analyses were used to identify the modifiable factors for MCI subtypes and predictors of cognitive decline,respectively.Results Among our participants,aMCI and VCIND demonstrated prevalence of 14.83%and 2.71%,respectively,and annual incidence(per 1000 person-years)of 69.6 and 10.6,respectively.The risk factor for aMCI was age,whereas its protective factors were high education level,tea consumption and physical activity.Moreover,VCIND risk factors were age,hypertension and depression.The presence of endocrine disease,cerebral trauma or hypertension was associated with a faster decline in cognition over 1 year.Conclusions MCI is a serious health problem in China that will only worsen as the population ages if no widespread interventions are implemented.Preventive strategies that promote brain activity and support healthy lifestyle choices are required.We identified modifiable factors for MCI in older individuals.The easy-to-adopt solutions such as tea consumption and physical activity can aid in preventing MCI.展开更多
Single-pixel imaging(SPI)is a prominent scattering media imaging technique that allows image transmission via one-dimensional detection under structured illumination,with applications spanning from long-range imaging ...Single-pixel imaging(SPI)is a prominent scattering media imaging technique that allows image transmission via one-dimensional detection under structured illumination,with applications spanning from long-range imaging to microscopy.Recent advancements leveraging deep learning(DL)have significantly improved SPI performance,especially at low compression ratios.However,most DL-based SPI methods proposed so far rely heavily on extensive labeled datasets for supervised training,which are often impractical in real-world scenarios.Here,we propose an unsupervised learningenabled label-free SPI method for resilient information transmission through unknown dynamic scattering media.Additionally,we introduce a physics-informed autoencoder framework to optimize encoding schemes,further enhancing image quality at low compression ratios.Simulation and experimental results demonstrate that high-efficiency data transmission with structural similarity exceeding 0.9 is achieved through challenging turbulent channels.Moreover,experiments demonstrate that in a 5 m underwater dynamic turbulent channel,USAF target imaging quality surpasses traditional methods by over 13 dB.The compressive encoded transmission of 720×720 resolution video exceeding 30 seconds with great fidelity is also successfully demonstrated.These preliminary results suggest that our proposed method opens up a new paradigm for resilient information transmission through unknown dynamic scattering media and holds potential for broader applications within many other scattering media imaging technologies.展开更多
基金funded by Guangzhou Huashang University(2024HSZD01,HS2023JYSZH01).
文摘Graph Neural Networks(GNNs),as a deep learning framework specifically designed for graph-structured data,have achieved deep representation learning of graph data through message passing mechanisms and have become a core technology in the field of graph analysis.However,current reviews on GNN models are mainly focused on smaller domains,and there is a lack of systematic reviews on the classification and applications of GNN models.This review systematically synthesizes the three canonical branches of GNN,Graph Convolutional Network(GCN),Graph Attention Network(GAT),and Graph Sampling Aggregation Network(GraphSAGE),and analyzes their integration pathways from both structural and feature perspectives.Drawing on representative studies,we identify three major integration patterns:cascaded fusion,where heterogeneous modules such as Convolutional Neural Network(CNN),Long Short-Term Memory(LSTM),and GraphSAGE are sequentially combined for hierarchical feature learning;parallel fusion,where multi-branch architectures jointly encode complementary graph features;and feature-level fusion,which employs concatenation,weighted summation,or attention-based gating to adaptively merge multi-source embeddings.Through these patterns,integrated GNNs achieve enhanced expressiveness,robustness,and scalability across domains including transportation,biomedicine,and cybersecurity.
基金Support by Sichuan Science and Technology Program[2023YFSY0026,2023YFH0004]Guangzhou Huashang University[2024HSZD01,HS2023JYSZH01].
文摘Robust teleoperation in image-guided interventions faces critical challenges from latency,deformation,and the quasi-periodic nature of physiological motion.This paper presents a fully integrated,latency-aware visual servoing system leveraging stereo vision,hand–eye calibration,and learning-based prediction for motion-compensated teleoperation.The system combines a calibrated binocular camera setup,dual robotic arms,and a predictive control loop incorporating Long Short-Term Memory(LSTM)and Temporal Convolutional Network(TCN)models.Through experiments using both in vivo and phantom datasets,we quantitatively assess the prediction accuracy and motion-compensation performance of both models.Results show that TCNs deliver more stable and precise tracking,especially on regular trajectories,while LSTMs exhibit robustness under quasi-periodic dynamics.By matching prediction horizons to system latency,the approach significantly reduces peak and steady-state tracking errors,demonstrating practical feasibility for deploying prediction-augmented servoing in teleoperated surgical.
基金Support by Sichuan Science and Technology Program[2023YFSY0026,2023YFH0004]Guangzhou Huashang University[2024HSZD01,HS2023JYSZH01].
文摘Graph neural networks(GNN)have shown strong performance in node classification tasks,yet most existing models rely on uniform or shared weight aggregation,lacking flexibility in modeling the varying strength of relationships among nodes.This paper proposes a novel graph coupling convolutional model that introduces an adaptive weighting mechanism to assign distinct importance to neighboring nodes based on their similarity to the central node.Unlike traditional methods,the proposed coupling strategy enhances the interpretability of node interactions while maintaining competitive classification performance.The model operates in the spatial domain,utilizing adjacency list structures for efficient convolution and addressing the limitations of weight sharing through a coupling-based similarity computation.Extensive experiments are conducted on five graph-structured datasets,including Cora,Citeseer,PubMed,Reddit,and BlogCatalog,as well as a custom topology dataset constructed from the Open University Learning Analytics Dataset(OULAD)educational platform.Results demonstrate that the proposed model achieves good classification accuracy,while significantly reducing training time through direct second-order neighbor fusion and data preprocessing.Moreover,analysis of neighborhood order reveals that considering third-order neighbors offers limited accuracy gains but introduces considerable computational overhead,confirming the efficiency of first-and second-order convolution in practical applications.Overall,the proposed graph coupling model offers a lightweight,interpretable,and effective framework for multi-label node classification in complex networks.
基金supported by the Major Project of Wuxi Municipal Health Commission[grant number:Z202406]the Jiangsu Commission of Health Program[grant number:M2024010]+3 种基金the National Key Research and Development Program[grant number:2022YFC3600600]the China Ministry of Science and Technology grants[grant number:2009BAI77B03]the Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support[grant number:20172029]the Innovative Research Team of High-level Local Universities in Shanghai[grant number:ZDCX20211201].
文摘Background As the population in China rapidly ages,the prevalence of mild cognitive impairment(MCI)is increasing considerably.However,the causes of MCI vary.The continued lack of understanding of the various subtypes of MCI impedes the implementation of effective measures to reduce the risk of advancing to more severe cognitive diseases.Aims To estimate the prevalence and incidence rates of two MCI subtypes—amnestic MCI(aMCI)and vascular cognitive impairment without dementia(VCIND)—and to determine modifiable factors for them among older individuals in a multiregional Chinese cohort.Method This 1-year longitudinal study surveyed a random sample of participants aged≥60 years from a large,community-dwelling cohort in China.Baseline lifestyle data were self-reported,while vascular and comorbid conditions were obtained from medical records and physical examinations.In total,3514 and 2051 individuals completed the baseline and 1-year follow-up assessments,respectively.Logistic and linear regression analyses were used to identify the modifiable factors for MCI subtypes and predictors of cognitive decline,respectively.Results Among our participants,aMCI and VCIND demonstrated prevalence of 14.83%and 2.71%,respectively,and annual incidence(per 1000 person-years)of 69.6 and 10.6,respectively.The risk factor for aMCI was age,whereas its protective factors were high education level,tea consumption and physical activity.Moreover,VCIND risk factors were age,hypertension and depression.The presence of endocrine disease,cerebral trauma or hypertension was associated with a faster decline in cognition over 1 year.Conclusions MCI is a serious health problem in China that will only worsen as the population ages if no widespread interventions are implemented.Preventive strategies that promote brain activity and support healthy lifestyle choices are required.We identified modifiable factors for MCI in older individuals.The easy-to-adopt solutions such as tea consumption and physical activity can aid in preventing MCI.
基金supported by the Natural Science Foundation of China Project(No.62525102).
文摘Single-pixel imaging(SPI)is a prominent scattering media imaging technique that allows image transmission via one-dimensional detection under structured illumination,with applications spanning from long-range imaging to microscopy.Recent advancements leveraging deep learning(DL)have significantly improved SPI performance,especially at low compression ratios.However,most DL-based SPI methods proposed so far rely heavily on extensive labeled datasets for supervised training,which are often impractical in real-world scenarios.Here,we propose an unsupervised learningenabled label-free SPI method for resilient information transmission through unknown dynamic scattering media.Additionally,we introduce a physics-informed autoencoder framework to optimize encoding schemes,further enhancing image quality at low compression ratios.Simulation and experimental results demonstrate that high-efficiency data transmission with structural similarity exceeding 0.9 is achieved through challenging turbulent channels.Moreover,experiments demonstrate that in a 5 m underwater dynamic turbulent channel,USAF target imaging quality surpasses traditional methods by over 13 dB.The compressive encoded transmission of 720×720 resolution video exceeding 30 seconds with great fidelity is also successfully demonstrated.These preliminary results suggest that our proposed method opens up a new paradigm for resilient information transmission through unknown dynamic scattering media and holds potential for broader applications within many other scattering media imaging technologies.
