UQGNN: Uncertainty Quantification of Graph Neural Networks for Multivariate Spatiotemporal Prediction

2025-08-11

AI Summary

Existing spatiotemporal forecasting models are predominantly deterministic, limiting uncertainty quantification; probabilistic alternatives often neglect dynamic interdependencies among heterogeneous urban phenomena. To address this, we propose UQGNN—the first unified framework integrating multivariate diffusion graph convolution with interaction-aware temporal networks, jointly modeling multivariate spatiotemporal dependencies and cross-phenomenon dynamic interactions while producing both predictive means and calibrated uncertainty estimates. Its core innovations include: (i) an interaction-aware spatiotemporal embedding module that captures coupled dynamics among heterogeneous urban processes; and (ii) a multivariate probabilistic forecasting module enabling end-to-end uncertainty inference. Extensive experiments on four real-world urban datasets—Shenzhen (two), New York, and Chicago—demonstrate that UQGNN consistently outperforms state-of-the-art methods in both prediction accuracy and uncertainty calibration, achieving an average 5% improvement on the Shenzhen dataset.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Spatiotemporal prediction plays a critical role in numerous real-world applications such as urban planning, transportation optimization, disaster response, and pandemic control. In recent years, researchers have made significant progress by developing advanced deep learning models for spatiotemporal prediction. However, most existing models are deterministic, i.e., predicting only the expected mean values without quantifying uncertainty, leading to potentially unreliable and inaccurate outcomes. While recent studies have introduced probabilistic models to quantify uncertainty, they typically focus on a single phenomenon (e.g., taxi, bike, crime, or traffic crashes), thereby neglecting the inherent correlations among heterogeneous urban phenomena. To address the research gap, we propose a novel Graph Neural Network with Uncertainty Quantification, termed UQGNN for multivariate spatiotemporal prediction. UQGNN introduces two key innovations: (i) an Interaction-aware Spatiotemporal Embedding Module that integrates a multivariate diffusion graph convolutional network and an interaction-aware temporal convolutional network to effectively capture complex spatial and temporal interaction patterns, and (ii) a multivariate probabilistic prediction module designed to estimate both expected mean values and associated uncertainties. Extensive experiments on four real-world multivariate spatiotemporal datasets from Shenzhen, New York City, and Chicago demonstrate that UQGNN consistently outperforms state-of-the-art baselines in both prediction accuracy and uncertainty quantification. For example, on the Shenzhen dataset, UQGNN achieves a 5% improvement in both prediction accuracy and uncertainty quantification.

Problem

• Quantifying uncertainty in multivariate spatiotemporal graph neural networks
• Capturing correlations among heterogeneous urban phenomena predictions
• Overcoming deterministic models' limitations in uncertainty estimation

Innovation

• Multivariate diffusion graph convolutional network captures spatial interactions
• Interaction-aware temporal convolutional network models temporal patterns
• Probabilistic prediction module estimates both mean values and uncertainties