Graph Convolutional Network (GCN) has been widely applied in transportation demand prediction due to its excellent ability capture non-Euclidean spatial dependence among station-level or regional demands. However, most of the existing research, graph convolution was implemented on a heuristically generated adjacency matrix, which could neither reflect real relationships stations accurately, nor multi-level demands adaptively. To cope with above problems, this paper provides novel...

Taxi and sharing bike bring great convenience to urban transportation. A lot of efforts have been made improve the efficiency taxi service or system by predicting next-period pick-up drop-off demand. Different from existing research, this paper is motivated following two facts: 1) From a micro view, an observed spatial demand at any time slot could be decomposed as combination many hidden bases; 2) macro multiple transportation demands are strongly correlated with each other, both spatially...

Recent studies have shown great promise in applying graph neural networks for multivariate time series forecasting, where the interactions of are described as a structure and variables represented nodes. Along this line, existing methods usually assume that (or adjacency matrix), which determines aggregation manner network, is fixed either by definition or self-learning. However, can be dynamic evolutionary real-world scenarios. Furthermore, quite different if they observed at scales. To...

Forecasting traffic flow is an important task in urban areas, and a large number of methods have been proposed for prediction. However, most the existing follow general technical route to aggregate historical information spatially temporally. In this paper, we propose different approach Our major motivation more effectively incorporate various intrinsic patterns real-world flows, such as fixed spatial distributions, topological correlations, temporal periodicity. Along line, novel...

Structural health monitoring system plays a vital role in smart management of civil engineering. A lot efforts have been motivated to improve data quality through mean, median values, or simple interpolation methods, which are low-precision and not fully reflected field conditions due the neglect strong spatio-temporal correlations borne by datasets thoughtless for various forms abnormal conditions. Along this line, article proposed an integrated framework augmentation structural using...

The purpose of this study is to explore what configurations dimensions corresponding environmental, social responsibility, governance (ESG) and firm contextual factors can lead the high-quality development state-owned enterprises (SOEs). A configuration analysis framework with six conditions including (ESG), innovation intensity, capital structure, size was constructed. Moreover, multi-stage qualitative comparative (QCA) approach conducted on a sample 692 annual observations SOEs from 2017...

Multivariate time series anomaly detection has great potentials in many practical applications. Extreme unbalanced training set and noise interference make it challenging to accurately capture the distribution of normal data then detect anomalies. Existing AutoEncoder(AE)-based approaches are lack effective regularization method specially designed for tasks thus easily overfitting while Generative Adversarial Network(GAN)-based mostly trained under hypothesis pollution-free set, which means...

Wound healing is a complex but precise physiological process. Howener, existing treatments are often difficult to meet the needs of different wound healing. With background that exogenous electrical stimulation (ES) has been proven be effective in regulating cell behavior, we constructed electroactive dressing derived from carbon nanotubes (CNT) by electrospinning technology. The scaffold moderate hydrophilicity, which benefits collecting effusion, adhering site, and safely removing....

Accurate prediction of the future mechanical behaviour underground structure is important for traffic system. However, existing works mostly just predicted one type property and failed to study influence different properties in tunnel. Besides, most them behaviours without considering external environment like temperature water pressure. In this paper, we propose a multi-task model named MSTNet which combines types indicators factors capturing temporal spatial characteristics Firstly,...

Patent classification aims to assign multiple International Classification (IPC) codes a given patent. Existing methods for automated patent primarily focus on analyzing the text descriptions of patents. However, apart from textual information, each is also associated with some assignees, and knowledge their previously applied patents can often be valuable accurate classification. Furthermore, hierarchical taxonomy defined by IPC system provides crucial contextual information enables models...

Mechanical analysis for the full face of tunnel structure is crucial to maintain stability, which a challenge in classical analytical solutions and data analysis. Along this line, study aims develop spatial deduction model obtain full-faced mechanical behaviors through integrating properties into pure data-driven model. The divided many parts reconstructed form matrix. Then, external load applied on field was considered tunnel. Based limited observed monitoring matrix results, double-driven...

Intelligent sensing, mechanism understanding, and the deterioration forecasting based on spatio–temporal big data not only promote safety of infrastructure but also indicate basic theory key technology for construction to turn intelligentization. The advancement underground space utilization has led development three characteristics (deep, big, clustered) that help shape a tridimensional urban layout. However, compared buildings bridges overground, diseases degradation occur are more...

Graph Convolutional Network (GCN) has been widely applied in transportation demand prediction due to its excellent ability capture non-Euclidean spatial dependence among station-level or regional demands. However, most of the existing research, graph convolution was implemented on a heuristically generated adjacency matrix, which could neither reflect real relationships stations accurately, nor multi-level demands adaptively. To cope with above problems, this paper provides novel...

In recent years, the application of graph neural networks (GNN) in multivariate time series forecasting has yielded remarkable achievements. The adjacency matrix which describes interactions among variables is dense and static most previous efforts no matter hand-crafted or self-learned. However, we argue that: 1) real-world scenario, could be dynamic evolving; 2) A sparse compact structure better reflect such interactions. Along this line, paper proposes a deep network based on GNN to...

Deep learning is one of the key topics research today. Researchers have found that deep algorithms developed rapidly in recent years, and this paper aims to apply practical problems civil engineering. Therefore, paper, we take images particles on a construction site segment using YOLO model, then calculate grading curves detection model. It use curve rapid convenient compared with traditional experimental methods; accuracy slightly flawed methods. With advancement hardware upgrading, method...

Abstract It is crucial to predict future mechanical behaviors for the prevention of structural disasters. Especially underground construction, are affected by multiple internal and external factors due complex conditions. Given that existing models fail take into account all accurate prediction time series simultaneously difficult using these models, this study proposed an improved model through autoencoder fused long‐ short‐term time‐series network driven mass number monitoring data. Then,...

Dynamic graph learning equips the edges with time attributes and allows multiple links between two nodes, which is a crucial technology for understanding evolving data scenarios like traffic prediction recommendation systems. Existing works obtain patterns mainly depending on most recent neighbor sequences. However, we argue that whether nodes will have interaction each other in future highly correlated same happened past. Only considering neighbors overlooks phenomenon of repeat behavior...

Exploiting self-supervised learning (SSL) to extract the universal representations of time series could not only capture natural properties but also offer huge help downstream tasks. Nevertheless, existing representation (TSRL) methods face challenges in attaining universality. Indeed, relying solely on one SSL strategy (either contrastive (CL) or generative) often fall short capturing rich semantic information for various Moreover, exhibit diverse distributions and inherent characteristics,...