SFADNet: Spatio-temporal Fused Graph based on Attention Decoupling Network for Traffic Prediction

2025-01-07

AI Summary

Existing traffic flow forecasting methods suffer from limited flexibility in spatiotemporal modeling and suboptimal prediction accuracy. To address these limitations, this paper proposes a multi-modal adaptive spatiotemporal graph learning framework. Its key contributions are: (1) an attention-driven time–space feature matrix disentanglement mechanism that explicitly isolates and models heterogeneous traffic patterns; (2) dynamic construction of pattern-specific adaptive fusion graphs, integrated with cross-attention and residual graph convolution for fine-grained spatiotemporal co-modeling; and (3) a joint optimization architecture for temporal modules and graph structures. Extensive experiments on four large-scale real-world datasets demonstrate that the proposed method consistently outperforms state-of-the-art approaches, achieving significant accuracy improvements in both short-term and medium-term forecasting tasks.

Technology Category

Planning, Routing, and Scheduling: Optimization of Spatio-temporal SystemsMachine Learning: Multimodal LearningComputer Vision: Multi-modal Vision

Application Category

Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMsSearch and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

In recent years, traffic flow prediction has played a crucial role in the management of intelligent transportation systems. However, traditional prediction methods are often limited by static spatial modeling, making it difficult to accurately capture the dynamic and complex relationships between time and space, thereby affecting prediction accuracy. This paper proposes an innovative traffic flow prediction network, SFADNet, which categorizes traffic flow into multiple traffic patterns based on temporal and spatial feature matrices. For each pattern, we construct an independent adaptive spatio-temporal fusion graph based on a cross-attention mechanism, employing residual graph convolution modules and time series modules to better capture dynamic spatio-temporal relationships under different fine-grained traffic patterns. Extensive experimental results demonstrate that SFADNet outperforms current state-of-the-art baselines across four large-scale datasets.

Problem

• Traffic Flow Prediction
• Time-Space Relationship
• Accuracy Improvement

Innovation

• SFADNet
• Temporal-Spatial Analysis
• Traffic Flow Prediction