LSTE-OATD: Direction-Driven Online Anomalous Trajectory Detection Using Learnable Spatio–Temporal Embeddings

ASTPKEFormer: Adaptive Spatiotemporal Prior Knowledge Embedding-Induced Transformers for Traffic Data Forecasting

ULSTM: Multi-Scale and Full-Level Temporal Consistency for Traffic Anomaly Detection

Urban traffic anomaly detection is essential for intelligent transportation systems, particularly in smart city environments where fast identification of abnormal events can improve road safety and traffic management. This work proposes a novel ULSTM-driven architecture that explicitly models temporal dependencies across consecutive traffic frames to achieve more stable and temporally coherent reconstructions. The proposed framework leverages sequential spatio-temporal representations to improve the distinction between normal traffic patterns and anomalous events. To further enhance reliability, we introduce a Hybrid Weighted Fusion strategy that synergistically combines structural, perceptual and pixel-wise metrics. The framework’s parameters are optimized using a Discrete Dirichlet Sampling approach, achieving a peak F1 Score of 70.28%. Evaluations were conducted on a manually curated traffic anomaly dataset with frame-level annotations. Experimental results demonstrate that the ULSTM framework significantly outperforms frame-independent generative models by suppressing high-frequency reconstruction noise, providing a robust solution for real-world smart city deployments. While highly effective in complex scenarios, the proposed framework is strictly applicable to highly dynamic traffic environments with active motion, as static background ensembles can degrade performance.

Federated spatio-temporal graph neural network for privacy-preserving vehicle trajectory prediction in autonomous driving

A Federated Spatio-Temporal Synchronous Dynamic Graph Neural Network (Federated STSDGNN) framework for privacy-preserving and adaptive trajectory prediction, which achieves approximately 23.5% lower RMSE compared to the centralized STSDGNN baseline in the conducted experiments, and has the potential to support future privacy-preserving V2X (Vehicle-to-Everything) applications.

Structure-Guided Spatiotemporal Attention Graph Neural Network for Traffic Flow Prediction

Deep spatiotemporal models integrating graph convolutions and attention mechanisms have demonstrated excellent performance in network-level traffic flow prediction, owing to their exceptional ability to capture complex spatiotemporal dependencies. Despite their predictive success, deployment of such models in safety-critical urban systems remains constrained by their inherent lack of transparency. Existing post-hoc diagnostic methods often struggle with spurious correlations and fail to unveil the intrinsic decision-making mechanisms governing traffic dynamics, resulting in suboptimal interpretability and limited operational trustworthiness. To address these challenges, this paper proposes the Structure-Guided Spatiotemporal Attention Graph Neural Network (SGSAN). Departing from traditional architectures that rely on unconstrained adaptive graphs, SGSAN explicitly learns a static Directed Dependency Graph (DDG) to identify the invariant macroscopic propagation paths of traffic states. We further introduce an InfoNCE-based soft-coupling mechanism that anchors the model's dynamic spatiotemporal attention to this structural prior, offering a mechanistic account of the model's decision-making process while ensuring robust forecasting by aligning attention-based reasoning with identified macroscopic dependencies and preventing over-reliance on ephemeral local noise. Furthermore, a decoupled two-stage optimization framework is developed to resolve the fundamental conflict between structural discovery and predictive error minimization. Extensive experiments on multiple real-world datasets demonstrate that SGSAN achieves state-of-the-art predictive accuracy while providing built-in interpretability that organically aligns with the physical logic of traffic networks.

A Hybrid Spatiotemporal Framework with Memory and Diffusion Convolution for Traffic Flow Prediction

Accurate traffic flow prediction is pivotal for intelligent transportation systems, yet it remains inherently challenging due to dynamic spatial correlations and long-range temporal dependencies. While existing forecasting paradigms predominantly rely on static, pre-defined graph structures, they often overlook the direct functional connections between non-adjacent time steps.This paper proposes a novel spatiotemporal forecasting framework, termed Memory-augmented Diffusion Convolutional LSTM network (MDC-LSTM), which integrates a MemBART-inspired memory mechanism, diffusion convolution, regional attention, and LSTM-based temporal modeling. By synthesizing these components with Long Short-Term Memory (LSTM) units, the proposed model effectively captures both localized spatial patterns and deep inter-temporal relationships. Empirical evaluations conducted on the METR-LA benchmark dataset demonstrate that our framework significantly enhances predictive accuracy across multiple metrics, consistently outperforming several state-of-the-art (SOTA) baselines and exhibiting strong robustness in long-term forecasting scenarios.

Physics-Informed Spatiotemporal Disentangling for Unsupervised Anomaly Detection in Earth Observation Sensing Nodes

Long-term health monitoring of unattended sensing nodes is essential for remote Earth observation networks (EONs); yet, it remains challenging because anomaly-induced response drifts often resemble genuine geophysical variations in both spectral and morphological characteristics. This ambiguity makes sensor degradation difficult to distinguish from valid observations, particularly when measurements are affected by regional spatiotemporal coupling and labeled fault data are unavailable. To address these issues, we propose a physics-informed and data-driven anomaly detection framework for EON sensing nodes. Multidomain complementary features describing amplitude, spectral, and phase behaviors are constructed to improve the separability between natural geophysical variability and sensor-induced distortions. Building on these features, a memory-enhanced Transformer-graph convolutional network (ME-TGCN) is developed to model spatiotemporal dependencies and disentangle node-specific abnormal responses from shared regional variations, while an external memory mechanism preserves long-term healthy operating patterns. Residual modeling errors are further compensated to improve normal-response estimation. Anomalies are then identified from the discrepancy between predicted and observed sensor responses through Mahalanobis-distance-based residual analysis with adaptive thresholding, enabling detection without labeled fault samples. Experiments on real-world EON datasets show that the proposed method outperforms representative baseline methods in both normal-state prediction and anomaly detection and can reliably track progressive sensor degradation and localized anomalies. Cross-regional transfer results further demonstrate its robustness, generalization capability, and practical value for large-scale unattended sensor network monitoring.