Uncertainty Quantification for Traffic Forecasting: A Unified Approach

08/11/2022
by   Weizhu Qian, et al.
0

Uncertainty is an essential consideration for time series forecasting tasks. In this work, we specifically focus on quantifying the uncertainty of traffic forecasting. To achieve this, we develop Deep Spatio-Temporal Uncertainty Quantification (DeepSTUQ), which can estimate both aleatoric and epistemic uncertainty. We first leverage a spatio-temporal model to model the complex spatio-temporal correlations of traffic data. Subsequently, two independent sub-neural networks maximizing the heterogeneous log-likelihood are developed to estimate aleatoric uncertainty. For estimating epistemic uncertainty, we combine the merits of variational inference and deep ensembling by integrating the Monte Carlo dropout and the Adaptive Weight Averaging re-training methods, respectively. Finally, we propose a post-processing calibration approach based on Temperature Scaling, which improves the model's generalization ability to estimate uncertainty. Extensive experiments are conducted on four public datasets, and the empirical results suggest that the proposed method outperforms state-of-the-art methods in terms of both point prediction and uncertainty quantification.

READ FULL TEXT

page 6

page 11

research
02/09/2023

A Benchmark on Uncertainty Quantification for Deep Learning Prognostics

Reliable uncertainty quantification on RUL prediction is crucial for inf...
research
12/09/2020

Uncertainty Intervals for Graph-based Spatio-Temporal Traffic Prediction

Many traffic prediction applications rely on uncertainty estimates inste...
research
08/11/2023

Uncertainty Quantification for Image-based Traffic Prediction across Cities

Despite the strong predictive performance of deep learning models for tr...
research
05/31/2023

DiffLoad: Uncertainty Quantification in Load Forecasting with Diffusion Model

Electrical load forecasting is of great significance for the decision ma...
research
02/06/2023

Dropout Injection at Test Time for Post Hoc Uncertainty Quantification in Neural Networks

Among Bayesian methods, Monte-Carlo dropout provides principled tools fo...
research
03/16/2023

Adaptive Modeling of Uncertainties for Traffic Forecasting

Deep neural networks (DNNs) have emerged as a dominant approach for deve...

Please sign up or login with your details

Forgot password? Click here to reset