Spatio-Temporal Self-Supervised Learning for Traffic Flow Prediction

Abstract

Robust prediction of citywide traffic flows at different time periods plays a crucial role in intelligent transportation systems. While previous work has made great efforts to model spatio-temporal correlations, existing methods still suffer from two key limitations: i) Most models collectively predict all regions' flows without accounting for spatial heterogeneity, i.e., different regions may have skewed traffic flow distributions. ii) These models fail to capture the temporal heterogeneity induced by time-varying traffic patterns, as they typically model temporal correlations with a shared parameterized space for all time periods. To tackle these challenges, we propose a novel Spatio-Temporal Self-Supervised Learning (ST-SSL) traffic prediction framework which enhances the traffic pattern representations to be reflective of both spatial and temporal heterogeneity, with auxiliary self-supervised learning paradigms. Specifically, our ST-SSL is built over an integrated module with temporal and spatial convolutions for encoding the information across space and time. To achieve the adaptive spatio-temporal self-supervised learning, our ST-SSL first performs the adaptive augmentation over the traffic flow graph data at both attribute- and structure-levels. On top of the augmented traffic graph, two SSL auxiliary tasks are constructed to supplement the main traffic prediction task with spatial and temporal heterogeneity-aware augmentation. Experiments on four benchmark datasets demonstrate that ST-SSL consistently outperforms various state-of-the-art baselines. Since spatio-temporal heterogeneity widely exists in practical datasets, the proposed framework may also cast light on other spatial-temporal applications. Model implementation is available at https://github.com/Echo-Ji/ST-SSL.

Traffic Prediction with Self-Supervised Learning: A Heterogeneity-Aware Model for Urban Traffic Flow Prediction Based on Self-Supervised Learning

“…Additionally, we included the backbone network ConvLSTM of the spatialtemporal encoder as a baseline to demonstrate the effectiveness of our self-supervised learning paradigms. To ensure fairness, we trained ConvLSTM and TPSSL with five different seeds, just like the baseline models whose results come from Ji et al [31].…”

“…Ji et al [30] adopted a self-supervised learning paradigm based on temporal continuity to examine the context information of traffic data, thereby better understanding and predicting the dynamic changes in traffic flow. Another study by Ji et al [31] proposed a contrastive learning-based traffic prediction framework and learned the representation of traffic data through auxiliary tasks to improve traffic prediction accuracy. Our approach differs from these studies because we spatially model traffic flow data as regular grids rather than as a graph.…”

HyFish: hydrological factor fusion for prediction of fishing effort distribution with VMS dataset

“…ST-SSL (Ji et al, 2023) concentrates on improving the representation of traffic patterns to accurately reflect spatial and temporal heterogeneity. It introduces a spatial-temporal selfsupervised learning framework specifically for traffic prediction.…”

Spatial-Temporal Graph Contrastive Learning for Urban Traffic Flow Forecasting

“…STGCL [21] devises four novel data augmentation approaches, and then demonstrates the effectiveness of joint learning schemes and graph-level contrasting for traffic forecasting through comprehensive experiments. ST-SSL [22] constructs spatial-temporal self-supervised auxiliary tasks based on adaptive augmented traffic graph to tackle the spatial-temporal heterogeneity of the urban network. For crime prediction, ST-HSL [40] designs a cross-region hypergraph and a dual-stage self-supervised learning paradigm to solve the challenge of sparse supervision signals.…”

“…For instance, to analyze the efficiency of the joint learning approach which simultaneously conducts forecasting and contrastive tasks, STGCL [21] introduces four effective data augmentation strategies to perturb inputs in terms of graph structure, frequency domain and time domain. ST-SSL [22] explores how to tackle the spatial-temporal heterogeneity with the self-supervised and contrastive learning techniques. Generally, contrastive learning methods utilize diverse data augmentation strategies to generate augmented views.…”