Abstract:To achieve accurate traffic forecasting, previous research has employed inner and outer aggregation for information aggregation, and attention mechanisms for heterogeneous spatiotemporal dependency learning, which results in inefficient model learning. While learning efficiency is critical due to the need for updating frequently the model to alleviate the impact of concept drift, limited work has focused on improving it. For efficient learning and accurate forecasting, this study proposes the dynamic spatiotemporal straight-flow network (DSTSFN). Breaking the aggregation paradigms employing both inner and outer aggregation, which may be redundant, the DSTSFN designs a straight-flow network that employs bipartite graphs to learn directly the dependencies between the source and target nodes for outer aggregation only. Instead of the attention mechanisms, the dynamic graphs/networks, which outdo static ones by possessing time-varying dependencies, are designed in the DSTSFN to distinguish the dependency heterogeneity, making the model relatively streamlined. Additionally, two learning strategies based on respectively the curriculum and transfer learning are developed to further improve the learning efficiency of the DSTSFN. Our study could be the first work designing the learning strategies for the multi-step traffic predictor based on dynamic spatiotemporal graphs. The learning efficiency and forecasting accuracy are demonstrated by experiments, which show that the DSTSFN can outperform not only the state-of-the-art (SOTA) predictor for accuracy by achieving a 2.27% improvement in accuracy and requiring only 8.98% of the average training time, but also the SOTA predictor for efficiency by achieving a 9.26% improvement in accuracy and requiring 91.68% of the average training time.

STLGRU: Spatio-Temporal Lightweight Graph GRU for Traffic Flow Prediction

Attention-Based Gated Recurrent Graph Convolutional Network for Short-Term Traffic Flow Forecasting

SGRU: A High-Performance Structured Gated Recurrent Unit for Traffic Flow Prediction

Global-Aware Enhanced Spatial-Temporal Graph Recurrent Networks: A New Framework For Traffic Flow Prediction

Spatial-Temporal Graph Attention Gated Recurrent Transformer Network for Traffic Flow Forecasting

Modeling Dynamic Traffic Flow As Visibility Graphs: A Network-Scale Prediction Framework for Lane-Level Traffic Flow Based on Lpr Data

Traffic flow prediction based on graph convolutional networks with a parallel attention network and stacked gate recurrent units

STG4Traffic: A Survey and Benchmark of Spatial-Temporal Graph Neural Networks for Traffic Prediction

STGRAT: A Spatio-Temporal Graph Attention Network for Traffic Forecasting

Reinforced Spatiotemporal Attentive Graph Neural Networks for Traffic Forecasting.

Adaptive spatial-temporal graph attention networks for traffic flow forecasting

STSD: Modeling Spatial Temporal Staticity and Dynamicity in Traffic Forecasting

STBGRN: A Traffic Prediction Model Based on Spatiotemporal Bidirectional Gated Recurrent Units and Graph Convolutional Residual Networks

Intelligent Traffic Flow Prediction Using Optimized GRU Model

Automated Dilated Spatio-Temporal Synchronous Graph Modeling for Traffic Prediction

A Graph Convolutional Stacked Temporal Attention Neural Network for Traffic Flow Forecasting

Dynamic Spatiotemporal Straight-Flow Network for Efficient Learning and Accurate Forecasting in Traffic

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

STGC-GNNs: A GNN-based Traffic Prediction Framework with a Spatial-Temporal Granger Causality Graph

TrafficStream: A Streaming Traffic Flow Forecasting Framework Based on Graph Neural Networks and Continual Learning

Attention-based Spatial-Temporal Graph Convolutional Recurrent Networks for Traffic Forecasting