spatiotemporal gnn

Traditional ST-GCNs employ单一 temporal modules—either CNNs or LSTMs—leading to insufficient capture of dynamic spatiotemporal patterns. To address this, we propose a plug-and-play hybrid temporal module that, for the first time, synergistically integrates CNNs and LSTMs within a unified co-temporal block. This design jointly models local temporal features and long-range dependencies. Through theoretical analysis and cross-dataset ablation studies, we systematically characterize the intrinsic relationship between temporal module architecture and representational capacity. Evaluated on standard spatiotemporal graph benchmarks—including NTU-RGB+D and PeMSD7—our method achieves significant improvements in prediction accuracy and cross-domain generalization. It consistently outperforms pure-CNN and pure-LSTM baselines in temporal representation learning. The proposed module establishes a reusable, principled design paradigm for temporal modeling in ST-GCNs, advancing both expressiveness and architectural flexibility.

Existing spatiotemporal graph models—such as GNNs and Transformers—typically adopt decoupled spatial and temporal modeling, limiting their ability to capture nontrivial structural dependencies inherent in graph topologies. To address this, we propose Cy2Mixer, a three-module architecture integrating temporal modeling, standard message passing, and a novel recurrent message-passing block (RMPB). The RMPB’s core innovation lies in explicitly constructing and aggregating cycle subgraphs to encode topological invariants; we theoretically prove its information complementarity with conventional message passing, thereby overcoming the limitations of spatiotemporal decoupling. Additionally, Cy2Mixer incorporates a gMLP backbone, gating mechanisms, and mathematically grounded topological representations. Extensive experiments on multiple spatiotemporal forecasting benchmarks demonstrate state-of-the-art performance, with significant improvements in prediction accuracy and generalization—particularly for traffic flow forecasting.

To address poor robustness in regional representation caused by strong noise and sparse labels in urban spatiotemporal graph data, this paper introduces the Spatiotemporal Heterogeneous Graph Neural Encoder (ST-HGAE), the first application of masked autoencoding to spatiotemporal graph learning. ST-HGAE jointly masks node features and graph structure, enabling generative self-supervised learning to automatically distill dynamic spatiotemporal dependencies. Its core innovations include a structure-aware masking strategy tailored for heterogeneous spatiotemporal graphs, and a dual reconstruction objective integrating node-level feature recovery with topology reconstruction. Evaluated on traffic flow, pedestrian flow, and crime prediction tasks, ST-HGAE consistently outperforms state-of-the-art methods—particularly under high noise levels and low label rates—while significantly enhancing modeling of dynamic spatial correlations among regions.

Existing model evaluation methods for spatiotemporal data—characterized by co-occurring missingness and heterogeneity, strong nonlinearity, and nonstationarity—lack interpretability and robustness. Method: We propose the first assumption-free, distribution-agnostic residual correlation diagnostic framework. It quantifies residual dependence structures across spatiotemporal dimensions via spatiotemporal graph modeling and asymptotically distribution-free autocorrelation statistics, enabling precise localization of local underfitting regions. Crucially, it imposes no prior assumptions on data distribution or underlying dynamics and natively supports interpretability assessment for sparse observations and nonlinear models—including spatiotemporal graph neural networks. Results: Extensive validation on synthetic and real-world datasets demonstrates that our framework accurately identifies performance-weak subregions, significantly enhancing the targeting and efficiency of model iteration.

Existing graph deep learning approaches for multivariate time series forecasting predominantly focus on architectural innovations, lacking systematic methodology studies on problem formalization, model design principles, and evaluation paradigms. This paper introduces the first unified methodology framework for graph-enhanced time series forecasting. It formally defines the forecasting problem by modeling dynamic inter-variable dependencies via learnable graph structures. It establishes principled design guidelines integrating graph neural networks (GNNs), spatiotemporal modeling, and deep temporal models (e.g., TCN and Informer variants). Furthermore, it proposes a reproducible evaluation protocol with built-in mechanisms to ensure interpretability and scalability. The framework shifts forecasting model development from empiricism toward first-principles reasoning, providing standardized guidance for relation-aware global modeling and articulating key open challenges. (149 words)

Existing graph neural networks are inherently limited to modeling pairwise relationships, struggling to effectively capture higher-order topological structures while suffering from rapidly escalating computational complexity as graph size grows. To address these limitations, this work proposes a simplicial complex–based spatiotemporal neural network that, for the first time, integrates simplicial complexes into spatiotemporal modeling. By leveraging spatiotemporal random walks on high-dimensional simplicial complexes and parallelized temporal convolutions, the proposed method transcends the pairwise interaction constraints of conventional graph neural networks. This approach significantly enhances the capacity to model higher-order topological dependencies in complex systems while maintaining computational efficiency.

This paper addresses the spatiotemporal forecasting problem of service traffic in distributed backend systems. We propose an end-to-end graph neural network–based modeling approach that innovatively represents service invocation dependencies as a dynamic directed graph. The method integrates multi-order graph convolutional layers—capturing topological dependencies—with gated recurrent units—modeling temporal evolution—to construct a lightweight spatiotemporal encoder. Trained via mean squared error minimization, the model achieves significant improvements over state-of-the-art baselines on a public microservice log dataset: average reductions of 12.7%–18.3% in MSE, RMSE, MAE, and MAPE. It maintains high accuracy and strong robustness across multi-step horizons (1–6 steps) and varying model depths. These results empirically validate the effectiveness of graph-structured abstraction and joint spatiotemporal modeling for service traffic forecasting.

Event camera data exhibits sparse and asynchronous characteristics, yet existing graph-based representations inadequately model spatiotemporal dynamics, limiting object detection performance. To address this, we propose a spatiotemporal multi-graph disentangled representation: spatially, global structure is modeled via B-spline basis functions; temporally, motion-vector-driven attention captures local dynamics. This joint design preserves data sparsity while enhancing spatiotemporal modeling fidelity. Crucially, our method replaces computationally expensive 3D convolutions with lightweight 2D convolutions fused with graph neural networks, enabling efficient asynchronous inference. Evaluated on Gen1 and eTraM benchmarks, our approach achieves over 6% improvement in detection accuracy, a 5× speedup in inference latency, and substantial reduction in parameter count—without increasing computational cost.

This work addresses the challenge in multivariate time series anomaly detection where over-generalized spatial structure modeling leads to erroneous reconstruction of anomalies and reduced recall. To mitigate this, we propose a prior-observation adversarial learning paradigm that jointly optimizes spatiotemporal dependency modeling. Our approach alternately learns an adjacency matrix as a structural prior in the spatial domain and employs a minimax adversarial mechanism to capture the discrepancy between this prior and data-driven observations, thereby significantly enhancing anomaly sensitivity along the temporal dimension and enabling, for the first time, channel-level anomaly localization. To facilitate systematic evaluation, we construct the first synthetic benchmark with precise channel-wise anomaly annotations. Extensive experiments demonstrate that our method achieves state-of-the-art performance on multiple public datasets as well as our newly introduced benchmark, excelling in both temporal detection and spatial localization tasks.

To address the challenges of modeling dynamic cross-client spatial dependencies and balancing privacy with predictive performance in traffic flow forecasting under federated learning, this paper proposes FedSTGD—a novel federated spatio-temporal graph learning framework. FedSTGD is the first to explicitly model dynamic spatial dependencies in a federated setting. It introduces a nonlinear graph computation decomposition mechanism and a node embedding enhancement module to decouple complex graph operations and strengthen local representation capability. Additionally, it establishes a lightweight server–client coordination protocol to enable efficient distributed spatio-temporal graph learning. Extensive experiments on four real-world traffic datasets demonstrate that FedSTGD consistently outperforms existing state-of-the-art methods in RMSE, MAE, and MAPE, achieving performance close to centralized training. Ablation studies validate the effectiveness of each component, while hyperparameter analysis confirms strong robustness.