检索结果-内蒙古大学图书馆

Identification of accident-injury type and bodypart factors from construction accident reports: A graph-based deep learning framework

引用

ADVANCED ENGINEERING INFORMATICS 2022年第0期54卷

作者： Pan, Xing Zhong, Botao Wang, Yuhang Shen, Luoxin Huazhong Univ Sci & Technol Natl Ctr Technol Innovat Digital Construction Wuhan Hubei Peoples R China Huazhong Univ Sci & Technol Sch Civil & Hydraul Engn Huazhong Peoples R China

Accident reports provide information to understand why and how events occur. learning from past accident reports is critical for preventing accidents or injuries in construction safety management. However, there are two issues: (1) manual analysis of such accident reports is time-consuming and labor-intensive;and (2) previous research mainly focused on analyzing the causal factors of accidents. Not much research concentrates on the injury effect in an accident and the influential relationship between accident cause and injury effect. To tackle this problem, a graph-based deep learning framework is proposed to identify accident-injury type and bodypart factors automatically to enable managers to make timely and better-informed decisions to prevent accidents and injuries for on-site safety. In this framework, a graph-based deep learning approach (specifically, the graph Convolutional Network) is developed to automatically classify accident reports labeled with accident_type and injury_type, whereas the traversal method is developed to identify the bodypart factors. To further intuitively visualize these safety risk factors (e.g., accident_type, injury_type, and bodypart factors), the co-occurrence networks are drawn to further intuitively reveal the interdependency in accident-injury and injury-bodypart types respectively. From the perspective of theoretical and practical contributions, the framework proposed in this study not only represents a substantial data-driven advancement in construction accident report classifi-cation and keyword extraction tasks, but also enables managers to get knowledge of construction safety per-formance (i.e., accident causes and injury effects) and further formulate corresponding strategies to prevent accidents and injuries in on-site safety management.

关键词： Construction safety management Accident reports graph-based deep learning Text classification Keyword extraction Safety risk factors

来源：评论

学校读者我要写书评

暂无评论

A comprehensive overview of graph neural network-based approaches to clustering for spatial transcriptomics

引用

COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL 2024年 23卷 106-128页

作者： Liu, Teng Fang, Zhao-Yu Zhang, Zongbo Yu, Yongxiang Li, Min Yin, Ming Zhu Chongqing Univ Chongqing Univ Three Gorges Hosp Clin Res Ctr CRC Clin Pathol Ctr CPCCanc Early Detect & Treatment Chongqing Peoples R China Chongqing Univ Chongqing Univ Three Gorges Hosp Translat Med Res Ctr TMRC Chongqing Peoples R China Chongqing Tech Innovat Ctr Qual Evaluat & Identica Chongqing Branch Chongqing Peoples R China Cent South Univ Sch Comp Sci & Engn Changsha 410083 Peoples R China Cent South Univ Hunan Prov Engn Res Ctr Intelligent Comp Biol & Me Changsha 410083 Peoples R China

Spatial transcriptomics technologies enable researchers to accurately quantify and localize messenger ribonucleic acid (mRNA) transcripts at a high resolution while preserving their spatial context. The identification of spatial domains, or the task of spatial clustering, plays a crucial role in investigating data on spatial transcriptomes. One promising approach for classifying spatial domains involves the use of graph neural networks (GNNs) by leveraging gene expressions, spatial locations, and histological images. This study provided a comprehensive overview of the most recent GNN-based methods of spatial clustering methods for the analysis of data on spatial transcriptomics. We extensively evaluated the performance of current methods on prevalent datasets of spatial transcriptomics by considering their accuracy of clustering, robustness, data stabilization, relevant requirements, computational efficiency, and memory use. To this end, we explored 60 clustering scenarios by extending the essential frameworks of spatial clustering for the selection of the GNNs, algorithms of downstream clustering, principal component analysis (PCA)-based reduction, and refined methods of correction. We comparatively analyzed the performance of the methods in terms of spatial clustering to identify their limitations and outline future directions of research in the field. Our survey yielded novel insights, and provided motivation for further investigating spatial transcriptomics.

关键词： Spatial transcriptome Cell type identification graph-based deep learning graph neural networks Survey

来源：评论

学校读者我要写书评

暂无评论

Survey on graph Neural Networks

引用

IEEE ACCESS 2024年 12卷 128816-128832页

作者： Gkarmpounis, Georgios Vranis, Christos Vretos, Nicholas Daras, Petros Ctr Res & Technol Hellas CERTH Informat Technol Inst ITI Thessaloniki 57001 Greece

graph Neural Networks (GNNs) have become a powerful tool in order to learn from graph-structured data. Their ability to capture complex relationships and dependencies within graph structures, allows them to have a great number of applications in various domains, including social network analysis, recommendation systems and drug discovery. The aim of this work is to provide a detailed overview of the models in graph Neural Networks and propose a new taxonomy of GNNs, including deep Generative Models for graphs as a distinct category. The works included were selected based on their impact, with recent related papers also considered.

关键词： Surveys graph neural networks Analytical models Filters Computational modeling Taxonomy deep learning graph neural networks (GNNs) taxonomy of GNNs graph-based deep learning

来源：评论

学校读者我要写书评

暂无评论

SpatialcoGCN: deconvolution and spatial information-aware simulation of spatial transcriptomics data via deep graph co-embedding

引用

BRIEFINGS IN BIOINFORMATICS 2024年第3期25卷 bbae130-bbae130页

作者： Yin, Wang Wan, You Zhou, Yuan Peking Univ Dept Biomed Informat Beijing Peoples R China Peking Univ Neurosci Res Inst Sch Basic Med Sci Dept Neurobiol Beijing Peoples R China Peking Univ Sch Basic Med Sci Dept Biomed Informat 38 Xueyuan Rd Beijing 100191 Peoples R China

Spatial transcriptomics (ST) data have emerged as a pivotal approach to comprehending the function and interplay of cells within intricate tissues. Nonetheless, analyses of ST data are restricted by the low spatial resolution and limited number of ribonucleic acid transcripts that can be detected with several popular ST techniques. In this study, we propose that both of the above issues can be significantly improved by introducing a deep graph co-embedding framework. First, we establish a self-supervised, co-graph convolution network-based deep learning model termed SpatialcoGCN, which leverages single-cell data to deconvolve the cell mixtures in spatial data. Evaluations of SpatialcoGCN on a series of simulated ST data and real ST datasets from human ductal carcinoma in situ, developing human heart and mouse brain suggest that SpatialcoGCN could outperform other state-of-the-art cell type deconvolution methods in estimating per-spot cell composition. Moreover, with competitive accuracy, SpatialcoGCN could also recover the spatial distribution of transcripts that are not detected by raw ST data. With a similar co-embedding framework, we further established a spatial information-aware ST data simulation method, SpatialcoGCN-Sim. SpatialcoGCN-Sim could generate simulated ST data with high similarity to real datasets. Together, our approaches provide efficient tools for studying the spatial organization of heterogeneous cells within complex tissues.

关键词： spatial transcriptomics cell type deconvolution graph-based deep learning spatial transcriptomics data simulation

来源：评论

学校读者我要写书评

暂无评论

Enabling smart curb management with spatiotemporal deep learning

引用

COMPUTERS ENVIRONMENT AND URBAN SYSTEMS 2023年 99卷

作者： Hao, Haiyan Wang, Yan Du, Lili Chen, Shigang Univ Florida Coll Design Construct & Planning Dept Urban & Reg Planning 1480 Inner Rd Gainesville FL 32611 USA Univ Florida Florida Inst Built Environm Resilience Coll Design Construct & Planning 1480 Inner Rd Gainesville FL 32611 USA Univ Florida Dept Urban & Reg Planning POB 115706 Gainesville FL 32611 USA Univ Florida Florida Inst Built Environm Resilience POB 115706 Gainesville FL 32611 USA Univ Florida Dept Civil & Coastal Engn Gainesville FL 32611 USA Univ Florida Dept Comp & Informat Sci & Engn Gainesville FL 32611 USA

Curb spaces are important assets to cities. They are often used by travelers to switch transportation means, visitors to access curbside properties, and municipalities to place roadside infrastructure. The promotion of multi-modal transportation and the emergence of new mobility services have complicated both curb environ-ments and their management. Consequently, some cities have started to explore new strategies for curb man-agement, but lacked the anticipation on how different curb regulations and built-environment features may collectively influence curb-use patterns across user groups. We make a step toward smart curb environment by proposing a graph-based deep learning approach, i.e., MultiGCN-LSTM, to predict diverse curb uses across time and space. We used two graph convolution layers and an LSTM layer to capture the spatial, temporal, and se-mantic dependencies between curb regulations, built-environment semantics, and diverse curb uses. Two place -specific models were developed separately for a medium-sized college town and a metropolitan in the U.S. The effectiveness of the proposed models was validated with ablation studies and demonstrated in three scenario experiments. The research contributes to smart curb management in the face of more diversified and intensified curb uses with new mobility services and emerging vehicular technologies.

关键词： Built environment Curb management graph-based deep learning Smart cities Scenario planning

来源：评论

学校读者我要写书评

暂无评论

Fast Traffic Engineering by Gradient Descent with Learned Differentiable Routing 18

Fast Traffic Engineering by Gradient Descent with Learned Di...

引用

18th International Conference on Network and Service Management (CNSM) - Intelligent Management of Disruptive Network Technologies and Services

作者： Rusek, Krzysztof Almasan, Paul Suarez-Varela, Jose Cholda, Piotr Barlet-Ros, Pere Cabellos-Aparicio, Albert AGH Univ Sci & Technol Inst Telecommun Krakow Poland Univ Politecn Cataluna Barcelona Neural Networking Ctr Barcelona Spain

ISBN: (纸本)9783903176515

Emerging applications such as the metaverse, telesurgery or cloud computing require increasingly complex operational demands on networks (e.g., ultra-reliable low latency). Likewise, the ever-faster traffic dynamics will demand network control mechanisms that can operate at short timescales (e.g., sub-minute). In this context, Traffic Engineering (TE) is a key component to efficiently control network traffic according to some performance goals (e.g., minimize network congestion). This paper presents Routing By Backprop (RBB), a novel TE method based on graph Neural Networks (GNN) and differentiable programming. Thanks to its internal GNN model, RBB builds an end-to-end differentiable function of the target TE problem (MinMaxLoad). This enables fast TE optimization via gradient descent. In our evaluation, we show the potential of RBB to optimize OSPF-based routing (similar to 25% of improvement with respect to default OSPF configurations). Moreover, we test the potential of RBB as an initializer of computationally-intensive TE solvers. The experimental results show promising prospects for accelerating this type of solvers and achieving efficient online TE optimization.

关键词： Traffic Engineering Routing Optimization graph-based deep learning Gradient Descent

来源：评论

学校读者我要写书评

暂无评论

Understanding and classification of innate immune response through weighted edge representation learning with dual hypergraph transformation

引用

RESULTS IN ENGINEERING 2025年 25卷

作者： Sakhamuri, Mallikharjuna Rao Henna, Shagufta Creedon, Leo Meehan, Kevin Mcafee, Marion Atlantic Technol Univ Dept Comp Donegal Ireland Atlantic Technol Univ Ctr Math Modelling & Intelligent Syst Hlth & Envir Letterkenny Ireland

Existing immunology approaches for modeling peptide-human leukocyte antigen (HLA) interactions predominantly rely on traditional machine learning and convolutional neural networks. However, these methods inadequately capture the underlying high-order relationships within peptide-HLA interactions and fail to represent the varying strengths of these interactions, which are crucial for accurately predicting immune responses. This limitation impedes the ability to model binding affinities effectively, which is essential for understanding T-cell activation and advancing immunotherapeutics. To address this challenge, we propose a Weighted Dual Hypergraph Transformation (WDHT) framework that extends conventional hypergraph learning by incorporating binding affinities as hyperedge weights and leveraging physicochemical properties as node features. Our approach models peptide-HLA interactions as a weighted hypergraph, where edges represent binding affinities and nodes encode physicochemical attributes such as hydrophobicity, charge, and structural tendencies. By applying WDHT, we construct a dual representation that transforms hyperedges into nodes and vice versa, enhancing the modeling of high-order interactions and improving the expressiveness of edge relationships. This transformation enables a more comprehensive capture of multi-way dependencies, which conventional hypergraph models fail to differentiate. We validate our approach using COVID-19 and dengue datasets, demonstrating significant improvements in classification accuracy and unsupervised learning performance. Our method achieves 93% accuracy on dengue data, outperforming conventional models (82-86%). It also improves recall scores to 85% on cell data, 92% on COVID-19 convalescent, and 91% on COVID-19 unexposed datasets, compared to traditional models' recall scores of 76-84%, 68-74%, and 75-90%. These results highlight WDHT's effectiveness in capturing peptide-HLA interaction dynamics, providing a robust framewo

关键词： Immune response Hypergraph learning Edge representation learning graph-based deep learning Molecular interactions

来源：评论

学校读者我要写书评

暂无评论

Fast Traffic Engineering by Gradient Descent with Learned Differentiable Routing 22

Fast Traffic Engineering by Gradient Descent with Learned Di...

引用

Proceedings of the 18th International Conference on Network and Service Management

作者： Krzysztof Rusek Paul Almasan José Suárez-Varela Piotr Chołda Pere Barlet-Ros Albert Cabellos-Aparicio AGH University of Science and Technology Universitat Politècnica de Catalunya Spain

ISBN: (纸本)9783903176515

Emerging applications such as the metaverse, telesurgery or cloud computing require increasingly complex operational demands on networks (e.g., ultra-reliable low latency). Likewise, the ever-faster traffic dynamics will demand network control mechanisms that can operate at short timescales (e.g., sub-minute). In this context, Traffic Engineering (TE) is a key component to efficiently control network traffic according to some performance goals (e.g., minimize network congestion).This paper presents Routing By Backprop (RBB), a novel TE method based on graph Neural Networks (GNN) and differentiable programming. Thanks to its internal GNN model, RBB builds an end-to-end differentiable function of the target TE problem (MinMaxLoad). This enables fast TE optimization via gradient descent. In our evaluation, we show the potential of RBB to optimize OSPF-based routing (≈25% of improvement with respect to default OSPF configurations). Moreover, we test the potential of RBB as an initializer of computationally-intensive TE solvers. The experimental results show promising prospects for accelerating this type of solvers and achieving efficient online TE optimization.

关键词： routing optimization traffic engineering gradient descent graph-based deep learning

来源：评论

学校读者我要写书评

暂无评论

deep Fusion of Brain Structure-Function in Mild Cognitive Impairment

引用

MEDICAL IMAGE ANALYSIS 2021年 72卷 102082-102082页

作者： Zhang, Lu Wang, Li Gao, Jean Risacher, Shannon L. Yan, Jingwen Li, Gang Liu, Tianming Zhu, Dajiang Univ Texas Arlington Dept Comp Sci & Engn Arlington TX 76019 USA Univ Texas Arlington Dept Math Arlington TX 76019 USA Indiana Univ Sch Med Dept Radiol & Imaging Sci Indianapolis IN 46202 USA Indiana Univ Sch Med Sch Informat & Comp Indianapolis IN 46202 USA Univ North Carolina Biomed Res Imaging Ctr Chapel Hill NC 27599 USA Univ North Carolina Dept Radiol Chapel Hill NC 27599 USA Univ Georgia Dept Comp Sci Cort Architecture Imaging & Discovery Lab Athens GA 30602 USA Univ Georgia Bioimaging Res Ctr Athens GA 30602 USA

Multimodal fusion of different types of neural image data provides an irreplaceable opportunity to take advantages of complementary cross-modal information that may only partially be contained in single modality. To jointly analyze multimodal data, deep neural networks can be especially useful because many studies have suggested that deep learning strategy is very efficient to reveal complex and non-linear relations buried in the data. However, most deep models, e.g., convolutional neural network and its numerous extensions, can only operate on regular Euclidean data like voxels in 3D MRI. The inter-related and hidden structures that beyond the grid neighbors, such as brain connectivity, may be over-looked. Moreover, how to effectively incorporate neuroscience knowledge into multimodal data fusion with a single deep framework is understudied. In this work, we developed a graph-based deep neural network to simultaneously model brain structure and function in Mild Cognitive Impairment (MCI): the topology of the graph is initialized using structural network (from diffusion MRI) and iteratively updated by incorporating functional information (from functional MRI) to maximize the capability of differenti-ating MCI patients from elderly normal controls. This resulted in a new connectome by exploring "deep relations" between brain structure and function in MCI patients and we named it as deep Brain Connec-tome. Though deep brain connectome is learned individually, it shows consistent patterns of alteration comparing to structural network at group level. With deep brain connectome, our developed deep model can achieve 92.7% classification accuracy on ADNI dataset. (c) 2021 Elsevier B.V. All rights reserved.

关键词： Brain structure and function fusion graph-based deep learning MCI

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：