Urban metro flow prediction is of great value for operation scheduling, passenger management and personal travel planning. However, the problem challenging. First, different stations, e.g. transfer stations non-transfer have unique traffic patterns. Second, it difficult to model complex spatio-temporal dynamic relation stations. To address these challenges, we develop a graph relational learning (STDGRL) predict urban station flow. propose node embedding representation module capture...

Abstract The transformer-based approach excels in long-term series forecasting. These models leverage stacking structures and self-attention mechanisms, enabling them to effectively model dependencies data. While some approaches prioritize sparse attention tackle the quadratic time complexity of self-attention, it can limit information utilization. We introduce a creative double-branch mechanism that simultaneously captures intricate both temporal variable perspectives. Moreover, we propose...

Accurate transmission line tension prediction is crucial for avoiding power grid suffering from serious lines ice coating, which a very challenging problem affected by multiple complex reasons, e.g., the massive sharp changes accompanying occurrence of cover, and non-negligible association with influencing factors (likes local geographic information future weather forecasts). In addition, spatial-temporal correlation among different time intervals leads to their contribution degrees...

Learning temporal dependencies among targets (TDT) benefits better time series forecasting, where refer to the predicted sequence. Although autoregressive methods model TDT recursively, they suffer from inefficient inference and error accumulation. We argue that integrating learning into non-autoregressive is essential for pursuing effective efficient forecasting. In this study, we introduce differencing approach represent propose a parameter-free plug-and-play solution through an...

Unsupervised feature selection (UFS) has recently gained attention for its effectiveness in processing unlabeled high-dimensional data. However, existing methods overlook the intrinsic causal mechanisms within data, resulting of irrelevant features and poor interpretability. Additionally, previous graph-based fail to account differing impacts non-causal constructing similarity graph, which leads false links generated graph. To address these issues, a novel UFS method, called Causally-Aware...

Accurate prediction of metro Origin-Destination (OD) flow is essential for the development intelligent transportation systems and effective urban traffic management. Existing approaches typically either predict passenger outflow departure stations or inflow destination stations. However, we argue that travelers generally have clearly defined arrival stations, making these OD pairs inherently interconnected. Consequently, considering as a unified entity more accurately reflects actual travel...

Weather Forecasting is an attractive challengeable task due to its influence on human life and complexity in atmospheric motion. Supported by massive historical observed time series data, the suitable for data-driven approaches, especially deep neural networks. Recently, Graph Neural Networks (GNNs) based methods have achieved excellent performance spatio-temporal forecasting. However, canonical GNNs-based only individually model local graph of meteorological variables per station or global...

Weather forecasting is an attractive and challenging task due to its influence on human life the complexity of atmospheric motion. Backed by abundance observed data, this suitable for data-driven approaches, especially deep learning technology. Many methods focus how establish a neural network extract temporal patterns meteorological variables while ignoring information interaction different between regions. In paper, we propose novel Hierarchical Spatio-temporal Graph Neural Network...

Deep learning (DL) has shown great potential in enhancing the performance of traditional numerical weather prediction (NWP) methods forecasting. Certain applications such as wind power generation desire more accurate predictions, especially local areas, which is challenging due to limited observations and complex dynamics. To this end, paper introduces a DL-based heterogeneous model named DeepWind for NWP correction, can simultaneously correct diverse variables across multiple stations. In...

Spatio-temporal forecasting of future values spatially correlated time series is important across many cyber-physical systems (CPS). Recent studies offer evidence that the use graph neural networks to capture latent correlations between holds a potential for enhanced forecasting. However, most existing methods rely on pre-defined or self-learning graphs, which are either static unintentionally dynamic, and thus cannot model time-varying exhibit trends periodicities caused by regularity...

Large-scale knowledge graphs are structured to represent real world facts, but they far from completeness. A number of completion methods have been developed fill missing facts. In this paper, a novel Sentence-RCNN embedding model is proposed for graph completion. This represents facts as sentences, so that it can precisely capture long-term dependencies, local structure information and translational features simultaneously. addition, we propose new method construct negative samples (CNS),...

Urban metro flow prediction is of great value for operation scheduling, passenger management and personal travel planning. However, it faces two main challenges. First, different stations, e.g. transfer stations non-transfer have unique traffic patterns. Second, challenging to model complex spatio-temporal dynamic relation stations. To address these challenges, we develop a graph relational learning (STDGRL) predict urban station flow. propose node embedding representation module capture the...