Dynamic Graph Filters Networks

A Gray-box Model for Multistep Traffic Forecasting

Conference Paper (2020)
Authors

Guopeng Li (Transport and Planning)

Victor Knoop (Transport and Planning)

J.W.C. van Lint (Transport and Planning)

Affiliation
Transport and Planning
Copyright
© 2020 G. Li, V.L. Knoop, J.W.C. van Lint
To reference this document use:
https://doi.org/10.1109/ITSC45102.2020.9294627
More Info
expand_more
Publication Year
2020
Language
English
Copyright
© 2020 G. Li, V.L. Knoop, J.W.C. van Lint
Affiliation
Transport and Planning
ISBN (print)
978-1-7281-4150-3
ISBN (electronic)
978-1-7281-4149-7
DOI:
https://doi.org/10.1109/ITSC45102.2020.9294627
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Short-term traffic forecasting is one of the key functions in Intelligent Transportation System (ITS). Recently, deep learning is drawing more attention in this field. However, how to develop a deep learning based traffic forecasting model that can dynamically extract explainable spatial correlations from traffic data is still a challenging issue. The difficulty mainly comes from the inconsistency between static model structures and the dynamic evolution of traffic conditions. To overcome this difficulty, we proposed a novel multistep speed forecasting model, Dynamic Graph Filters Networks (DGFN). The major contribution is that the regular pixel-wise dynamic convolution is extended to graph topology. DGFN has a simple recurrent cell structure where local area-wide graph convolutional kernels are dynamically computed from varying inputs. Experiments on ring freeways show that DGFN is able to precisely predict short-term evolution of traffic speed. Furthermore, we theoretically explain why DGFN is not a pure “black-box”, but a “gray-box” model that actually reduces entangled spatial and temporal features into one component representing dynamic spatial correlations. It permits tracking real-time interactions among adjacent links. DGFN has the potential to relate trained parameters in deep learning models with physical traffic variables.

Files

09294627.pdf
(pdf | 1.73 Mb)
- Embargo expired in 24-06-2021
License info not available