DSTGCN: Dynamic Spatial-Temporal Graph Convolutional Network for Traffic Prediction

With the continuous development of smart cities, intelligent transportation systems (ITSs) have ushered in many breakthroughs and upgrades. As a solid foundation for an ITS, traffic flow prediction effectively helps the city to better manage intricate traffic flow. However, existing traffic flow prediction methods such as temporal graph convolutional networks(T-GCNs) ignore the dissimilarities between lanes. Thus, they cannot provide more specific information regarding predictions such as dynamic changes in traffic flow direction and deeper lane relationships. With the upgrading of intersection sensors, more and more intersection lanes are equipped with intersection sensors to detect vehicle information all day long. These spatio-temporal data help researchers refine the focus of traffic prediction research down to the lane level. More accurate and detailed data mean that it is more difficult to mine the spatio-temporal correlations between data, and modeling heterogeneous data becomes more challenging. In order to deal with these problems, we propose a heterogeneous graph convolution model based on dynamic graph generation. The model consists of three components. The internal graph convolution network captures the real-time spatial dependency between lanes in terms of generated dynamic graphs. The external heterogeneous data fusion network comprehensively considers other parameters such as lane speed, lane occupancy, and weather conditions. The codec neural network utilizes a temporal attention mechanism to capture the deep temporal dependency. We test the performance of this model based on two real-world datasets, and extensive comparative experiments indicate that the proposed heterogeneous graph convolution model can improve the prediction accuracy.

Dynamic multi-scale spatial-temporal graph convolutional network for traffic flow prediction

Graph Convolutional Networks (GCNs) have been widely used for collaborative filtering, due to their effectiveness in exploiting high-order collaborative signals. However, two issues have not been well addressed by existing studies. First, usually only one kind of information is utilized, i.e., user preference in user-item graphs or item dependency in item-item graphs. Second, they usually adopt static graphs, which cannot retain the temporal evolution of the information. These can limit the recommendation quality. To address these limitations, we propose to mine three kinds of information (user preference, item dependency, and user behavior similarity) and their temporal evolution by constructing multiple discrete dynamic heterogeneous graphs (i.e., a user-item dynamic graph, an item-item dynamic graph, and a user-subseq dynamic graph) from interaction data. A novel network (PDGCN) is proposed to learn the representations of users and items in these dynamic graphs. Moreover, we designed a structural neighbor aggregation module with novel pooling and convolution operations to aggregate the features of structural neighbors. We also design a temporal neighbor aggregation module based on self-attention mechanism to aggregate the features of temporal neighbors. We conduct extensive experiments on four real-world datasets. The results indicate that our approach outperforms several competing methods in terms of Hit Ratio (HR) and Normalized Discounted Cumulative Gain (NDCG). Dynamic graphs are also shown to be effective in improving recommendation performance.

Traffic prediction is a fundamental operation in real-time traffic analysis. A precise prediction of traffic condition can benefit both road users and traffic management agencies. However, since road traffic is decided by multiple static and dynamic factors, traffic prediction is still a challenging task. As the core indicator of traffic condition, many works focus on traffic speed prediction using time-series forecasting approaches. Although current methods take into account the static road topology while modelling, they fail to consider (1) the semantic closeness between road components and (2) congestion caused by upstream/downstream traffic propagation. In this paper, we introduce a Spatial-Temporal Dynamic Graph Network using JS-Graph, which considers both static road features and dynamic traffic flows when forecasting. Specifically, we first propose a data-driven ‘JS-Graph’ method that describes the semantic similarity between road nodes. It models the complex spatial correlations that cannot be captured by the traditional spatial adjacency graph. Secondly, we design a dynamic graph attention network that considers the traffic dynamics that happened in previous time slices when predicting the current one to capture the congestion propagation phenomena. Extensive experiments conducted on real-world datasets show that our proposed method is significantly better than baselines. KeywordsGraph neural networkSpatial-temporal data analysisTime series forecast

The graph neural network (GNN) has emerged as an effective way to mine relationships between data due to its powerful modeling capability for graph structure data, and it has garnered significant attention from researchers for intelligent fault diagnosis tasks. However, the adjacency matrix of most GNN models with deep architecture is always fixed during the aggregation process, and the edge connection relationship cannot be adaptively adjusted, which limits their performance for feature representation. Moreover, for few-shot diagnosis scenarios, the generalization performance of deep GNN models will be further degraded due to fixed receptive fields and limited training samples. To address these issues, a deep adaptively dynamic edge graph convolution network (DADE-GCN) with attention weight and a high-dimension affinity feature graph is proposed. First, a deep adaptively dynamic edge graph convolutional module with attention weight (DADE-GCNWAW) is developed to dynamically adjust the receptive field in different graph convolution layers. Subsequently, the output features of different layers are fused by a self-attention mechanism. Second, to overcome the effect of the time-shift problem existing in vibration signals and capture accurate interdependencies between data, a high-dimension affinity feature graph construction method is proposed to construct graph structure data. The effectiveness of the proposed method is quantitatively verified by two rotating machinery datasets, which indicate that the proposed DADE-GCN model can achieve average diagnosis accuracies of 98.80% in both the two few-shot diagnosis tasks, which outperform several state-of-the-art recognition methods.

Dynamic traffic correlations based spatio-temporal graph convolutional network for urban traffic prediction

DyHDGE: Dynamic heterogeneous transaction graph embedding for safety-centric fraud detection in financial scenarios

Short-term passenger flow prediction is crucial for urban traffic management and crowd control. However, traditional graph neural network methods usually use static adjacency matrices in urban rail transit prediction, ignoring the dynamic changes of inter-station correlations throughout the day. To address this challenge, we propose a new model called Dynamic Multigraph Fusion Spatio-Temporal Graph Convolutional Network (DMSTGCN). This model uses Tucker decomposition to construct a time-varying dynamic graph, capturing the spatial correlations of passenger flow at different time intervals and overcoming the limitations of static matrices in previous studies while effectively mitigating overfitting. Three predefined graphs, namely the physical structure graph, the passenger flow similarity graph and the OD correlation graph, are also integrated to form the fusion graph convolution module, which further enriches the multivariate spatial correlation of passenger flows. Meanwhile, the causal temporal multi-head self-attention mechanism extracts temporal correlations, working with the fusion graph convolution module in a spatio-temporal learner for multilayer spatio-temporal feature extraction. Finally, experiments on three real-world datasets prove the excellence of DMSTGCN and offer interpretable dynamic spatial relationships.

In the intelligent transportation system, traffic forecasting, which is generally characterized as a graph spatial-temporal prediction task, plays a crucial role. It is challenging to generate reliable forecast results due to the complexity of traffic topological information and the inherent uncertainty of road traffic circumstances. Existing works generally focus on modeling spatial dependency on static graph structures, but ignore dynamic relations between road segments and cannot extract long-range traffic dependencies in spatial-temporal domains. To bridge the above gaps, we present a novel framework, called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dynamic Multi-Hop Graph Attention Network</i> (DMGAN). Specifically, we leverage dynamic graph modeling to capture time-varying relations across road sections and introduce the multi-hop operation in each message propagation layer to extract long-range spatial dependency. Meanwhile, we develop a fusion-attention module, preserving both local and global hidden layer outputs of the encoder, to capture both long- and short-term temporal dependencies jointly. In this way, our method can fully model complex time-varying traffic topology information and capture the internal patterns of traffic series by integrating dynamic graph structure and temporal attention component. DGMAN achieves state-of-the-art performance in three metrics, as demonstrated by experimental findings on four real-world public traffic datasets, METR-LA, PEMS-BAY, PEMS03, and PEMS07. This code and data are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/EEHITer/2022-TKDE-DMGAN-Pytorch/tree/main</uri> for reproducibility and further studies.

TodyNet: Temporal dynamic graph neural network for multivariate time series classification

Accurate traffic flow prediction is vital for intelligent transportation systems, yet strong spatiotemporal coupling and multi-scale dynamics make modelling difficult. Existing methods often rely on static adjacency and short input windows, limiting adaptation to time-varying spatial relations and long-term patterns. To address these issues, we propose the Pre-trained Trend-aware Dynamic Graph Convolutional Network (PT-TDGCN), a two-stage framework. In the pre-training stage, a Transformer-based masked autoencoder learns segment-level temporal representations from historical sequences. In the prediction stage, three designs are integrated: (1) dynamic graph learning parameterized by tensor decomposition; (2) convolutional trend-aware attention that adds 1D convolutions to capture local trends while preserving global context; and (3) spatial graph convolution combined with lightweight fusion projection for aligning pre-trained, spatial, and temporal representations. Extensive experiments on four real-world datasets demonstrated that PT-TDGCN consistently outperformed 14 baseline models, achieving superior predictive accuracy and robustness.

Sequential recommendation system's goal is to predict users' next actions based on their historical behavior sequences. As we know, more recent items have a larger impact than the previous ones. Meanwhile, modeling users' current interests are challenging. Knowledge graph (KG) contains a vast of information, which can help us capture users' interests by propagating their interactions. In this paper, we propose an Attentive Dynamic Knowledge Graph Networks (ADKGN) for sequential recommendation, which includes an embedding module, an attentive dynamic knowledge graph module, a sequential process module and a prediction module. Specifically, embedding module learns initial item embedding vectors by combing latent features and sequential features. For users' recent interacted items, attentive dynamic knowledge graph module learns dynamic weights for each pretrained KG embedding, and then utilizes top-k layer and parallel-based aggregation layer to effectively aggregate useful information from multi-hop neighbors. Sequential process module combines a user's history interactions and processed current interactions, and employs sequential models over them to get final user representations. Prediction module predicts the clicking probability using final user representation and target item representation. We conduct experiments on a public dataset, finding that ADKGN significantly outperforms state-of-the-art solutions.

Dynamic gesture recognition based on skeletal data has garnered significant attention with the rise of graph convolutional networks (GCNs). Existing methods typically calculate dependencies between joints and utilize spatio-temporal attention features. However, they often rely on joint topological features of limited spatial extent and short-time features, making it challenging to extract intra-frame spatial features and long-term inter-frame temporal features. To address this, we propose a new GCN architecture for dynamic hand gesture recognition, called a spatio-temporal dynamic attention graph convolutional network (STDA-GCN). This model employs dynamic attention spatial graph convolution, enhancing spatial feature extraction capabilities while reducing computational complexity through improved cross-channel information interaction. Additionally, a salient location channel attention mechanism is integrated between spatio-temporal convolutions to extract useful spatial features and avoid redundancy. Finally, dynamic multi-scale temporal convolution is used to extract richer inter-frame gesture features, effectively capturing information across various time scales. Evaluations on the SHREC’17 Track and DHG-14/28 benchmark datasets show that our model achieves 97.14% and 95.84% accuracy, respectively. These results demonstrate the superior performance of STDA-GCN in dynamic gesture recognition tasks.

Electricity theft is a critical issue in modern power systems, leading to financial losses and reduced reliability of electricity distribution networks. The growing use of Advanced Metering Infrastructure (AMI) has enabled smart meters to collect detailed consumption data, offering new opportunities to detect abnormal usage patterns. However, many existing detection methods rely only on individual load curves and fail to capture temporal dependencies, periodic behaviors, and hidden correlations within the data. To overcome these limitations, this study proposes a Dynamic Generative Residual Graph Convolutional Neural Network (DG RGCN). The model constructs dynamic graphs by updating adjacency matrices during training, allowing it to represent evolving relationships among electricity users. To address the challenge of imbalanced datasets, SMOTE oversampling and class weight adjustment are applied, ensuring improved detection accuracy. The proposed approach is evaluated on the SGCC smart meter dataset, and experimental results demonstrate that DG RGCN significantly outperforms traditional machine learning and deep learning models. By combining dynamic graph learning with residual connections, the model provides a practical and effective solution for electricity theft detection, enhancing both security and efficiency in smart grid systems.

With the development of precision sensing instruments and data storage devices, the fusion of multi-sensor data in gearbox fault diagnosis has attracted much attention. However, existing methods have difficulty in capturing the local temporal dependencies of multi-sensor monitoring information, and the inescapable noise severely decreases the accuracy of multi-sensor information fusion diagnosis. To address these issues, this paper proposes a fault diagnosis method based on dynamic graph convolutional neural networks and hard threshold denoising. Firstly, considering that the relationships between monitoring data from different sensors change over time, a dynamic graph structure is adopted to model the temporal dependencies of multi-sensor data, and, further, a graph convolutional neural network is constructed to achieve the interaction and feature extraction of temporal information from multi-sensor data. Secondly, to avoid the influence of noise in practical engineering, a hard threshold denoising strategy is designed, and a learnable hard threshold denoising layer is embedded into the graph neural network. Experimental fault datasets from two typical gearbox fault test benches under environmental noise are used to verify the effectiveness of the proposed method in gearbox fault diagnosis. The experimental results show that the proposed DDGCN method achieves an average diagnostic accuracy of up to 99.7% under different levels of environmental noise, demonstrating good noise resistance.