Spatio-Temporal Meta Contrastive Learning (2023-10-21T00:00:00.000000Z)

TL;DR

This work proposes a new spatio-temporal contrastive learning (CL4ST) framework to encode robust and generalizable STG representations via the STG augmentation paradigm and designs the meta view generator to automatically construct node and edge augmentation views for each disentangled spatial and temporal graph.

Abstract

Spatio-temporal prediction is crucial in numerous real-world applications, including traffic forecasting and crime prediction, which aim to improve public transportation and safety management. Many state-of-the-art models demonstrate the strong capability of spatio-temporal graph neural networks (STGNN) to capture complex spatio-temporal correlations. However, despite their effectiveness, existing approaches do not adequately address several key challenges. Data quality issues, such as data scarcity and sparsity, lead to data noise and a lack of supervised signals, which significantly limit the performance of STGNN. Although recent STGNN models with contrastive learning aim to address these challenges, most of them use pre-defined augmentation strategies that heavily depend on manual design and cannot be customized for different Spatio-Temporal Graph (STG) scenarios. To tackle these challenges, we propose a new spatio-temporal contrastive learning (CL4ST) framework to encode robust and generalizable STG representations via the STG augmentation paradigm. Specifically, we design the meta view generator to automatically construct node and edge augmentation views for each disentangled spatial and temporal graph in a data-driven manner. The meta view generator employs meta networks with parameterized generative model to customize the augmentations for each input. This personalizes the augmentation strategies for every STG and endows the learning framework with spatio-temporal-aware information. Additionally, we integrate a unified spatio-temporal graph attention network with the proposed meta view generator and two-branch graph contrastive learning paradigms. Extensive experiments demonstrate that our CL4ST significantly improves performance over various state-of-the-art baselines in traffic and crime prediction. Our model implementation is available at the link: https://github.com/HKUDS/CL4ST.

References56 items

Automated Spatio-Temporal Graph Contrastive Learning

MSDR: Multi-Step Dependency Relation Networks for Spatial Temporal Forecasting

Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting

CausalGNN: Causal-Based Graph Neural Networks for Spatio-Temporal Epidemic Forecasting

Spatial-Temporal Hypergraph Self-Supervised Learning for Crime Prediction

ST-GSP: Spatial-Temporal Global Semantic Representation Learning for Urban Flow Prediction

Graph Neural Controlled Differential Equations for Traffic Forecasting

AutoGCL: Automated Graph Contrastive Learning via Learnable View Generators

When do contrastive learning signals help spatio-temporal graph forecasting?

DL-Traff: Survey and Benchmark of Deep Learning Models for Urban Traffic Prediction

Dynamic and Multi-faceted Spatio-temporal Deep Learning for Traffic Speed Forecasting

Spatial-Temporal Sequential Hypergraph Network for Crime Prediction with Dynamic Multiplex Relation Learning

Spatial-Temporal Graph ODE Networks for Traffic Flow Forecasting

Graph Contrastive Learning Automated

Adversarial Graph Augmentation to Improve Graph Contrastive Learning

Z-GCNETs: Time Zigzags at Graph Convolutional Networks for Time Series Forecasting

CLEAR: Contrastive Learning for Sentence Representation

Spatial-Temporal Fusion Graph Neural Networks for Traffic Flow Forecasting

Spectral Temporal Graph Neural Network for Multivariate Time-series Forecasting

Graph Contrastive Learning with Augmentations

Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting

LSGCN: Long Short-Term Traffic Prediction with Graph Convolutional Networks

Contrastive Multi-View Representation Learning on Graphs

Hierarchically Structured Transformer Networks for Fine-Grained Spatial Event Forecasting

Spatial-Temporal Synchronous Graph Convolutional Networks: A New Framework for Spatial-Temporal Network Data Forecasting

A Simple Framework for Contrastive Learning of Visual Representations

GMAN: A Graph Multi-Attention Network for Traffic Prediction

DSANet: Dual Self-Attention Network for Multivariate Time Series Forecasting

Urban Traffic Prediction from Spatio-Temporal Data Using Deep Meta Learning

Attention Based Spatial-Temporal Graph Convolutional Networks for Traffic Flow Forecasting

Graph WaveNet for Deep Spatial-Temporal Graph Modeling

UrbanFM: Inferring Fine-Grained Urban Flows

DeepCrime: Attentive Hierarchical Recurrent Networks for Crime Prediction

How Powerful are Graph Neural Networks?

Deep Graph Infomax

Revisiting Spatial-Temporal Similarity: A Deep Learning Framework for Traffic Prediction

Deep Multi-View Spatial-Temporal Network for Taxi Demand Prediction

Graph Attention Networks

Spatio-temporal Graph Convolutional Neural Network: A Deep Learning Framework for Traffic Forecasting

Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting

Categorical Reparameterization with Gumbel-Softmax

The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables

Deep Spatio-Temporal Residual Networks for Citywide Crowd Flows Prediction

Semi-Supervised Classification with Graph Convolutional Networks

Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

Utilizing Real-World Transportation Data for Accurate Traffic Prediction

LIBSVM: A library for support vector machines

TAMP-S2GCNets: Coupling Time-Aware Multipersistence Knowledge Representation with Spatio-Supra Graph Convolutional Networks for Time-Series Forecasting

AUTO-ENCODING VARIATIONAL BAYES

ConnectingtheDots:MultivariateTimeSeriesForecastingwith GraphNeuralNetworks

ExaminingCOVID-19ForecastingusingSpatio-Temporal Graph Neural Networks

Freeway Performance Measurement System: Mining Loop Detector Data

On robust estimation of the location parameter

2022. Towards Spatio-Temporal Aware Traffic Time Series Forecasting-Full Version

2022. FOGS: First-Order Gradient Supervision with Learning-based Graph for Traffic Flow Forecasting

2023. Dis-entangled Contrastive Collaborative Filtering