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 and effectively captures both localized spatial patterns and deep inter-temporal relationships.
Abstract
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.
TETRA is proposed, a hybrid spatio-temporal traffic forecasting model that integrates Graph Convolutional Networks (GCNs) with Extended Long Short-Term Memory (xLSTM) to capture complex multi-timescale temporal patterns, including congestion propagation and delayed recovery dynamics, which are not well represented by conventional recurrent models.
Norman Bereczki, Vilmos Simon· International Journal of Int...· 0 citations
This framework introduces an adaptive graph learning module that dynamically infers meaningful connectivity relationships among traffic sensors—not relying on fixed geographic or distance-based assumptions—but instead leveraging real-time traffic correlations and node-level embeddings, enabling effective modeling of both localized spatial interactions and multi-scale temporal dependencies across varying prediction horizons.
Zhengxu Luan, Huan Wang, Miaobowen Wang et al.· Computers and artificial int...· 0 citations
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 novel method called adaptive diffused spatiotemporal graph convolution network (ADSTGCN) is proposed for accurate traffic flow prediction and achieves superior performance compared to other state-of-the-art methods.
Xiao Luo, Shanshan Wang, Shaobao Li et al.· Journal of Transportation En...· 0 citations
Experimental results on the public PEMS04 and PEMS08 datasets demonstrate that the proposed ESDG-ALSTM model significantly improves forecasting accuracy, confirming that ESDG-ALSTM is more sensitive to abrupt events and multimodal evolution patterns and can effectively enhance prediction performance in complex traffic flow scenarios.
Guozheng Li, Baijing Wu, Ke Gao et al.· Frontiers of Computer Scienc...· 0 citations
Urban traffic flow is difficult to forecast accurately because its evolution is non-linear and governed by dependencies that operate over different spatial and temporal ranges. This paper introduces the Multi-Perspective Spatio-Temporal Feature Fusion Model (MPSTFFM) to describe these dependencies through complementary views. The temporal signal is separated into a slowly varying trend and a residual fluctuation, while the spatial structure is represented by four graphs: first-order adjacency, second-order in-degree, second-order out-degree, and a data-adaptive graph. These graphs respectively encode physical road connectivity, common inflow sources, common outflow destinations, and latent spatial associations. Whereas the first three are constructed from the known network topology, the adaptive graph is learned together with the prediction model and can therefore identify correlations not expressed by physical links. Within each spatio-temporal view, self-attention captures dependencies over long ranges, and convolutional operations extract local patterns. The features learned from all views are subsequently fused into a high-dimensional representation used to predict future flow. Experiments on real-world datasets compare MPSTFFM with twelve methods published during the preceding five years. On these benchmarks MPSTFFM outperforms every baseline, lowering the average MAE, RMSE, and MAPE across the four datasets by 13.04%, 5.28%, and 9.59%, respectively, relative to the best baseline on each one.
A. Marakhimov, Rustem Jalelov, J.K. Kudaybergenov et al.· Italian National Conference...· 0 citations