Graph Pyramid Autoformer for Long- Term Traffic Forecasting

Weiheng Zhong; Tanwi Mallick; Jane MacFarlane; Hadi Meidani; Prasanna Balaprakash

doi:10.1109/ICMLA58977.2023.00060

Graph Pyramid Autoformer for Long- Term Traffic Forecasting

Weiheng Zhong
, Tanwi Mallick
, Jane MacFarlane
, Hadi Meidani
, Prasanna Balaprakash

CCSD Planning, Partnerships, Operations

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Accurate traffic forecasting is vital to an intelligent transportation system. Although many deep learning models have achieved state-of-art performance for short-term traffic forecasting of up to 1 hour, long-term traffic forecasting that spans multiple hours remains a major challenge. To that end, we develop Graph Pyramid Autoformer (GPA), an attention-based spatial-temporal graph neural network that uses a novel pyramid autocorrelation attention mechanism. It enables learning from long temporal sequences on graphs and improves long-term traffic forecasting accuracy. We demonstrate the efficacy of the GPA using two benchmark traffic datasets: Los Angeles' METR-LA and the Bay Area's PEMS-BAY. Notably, our model has outperformed a range of existing state-of-the-art methods, delivering up to a 25 % improvement in the accuracy of long-term traffic forecasts. Our code is available at: https://github.com/WeiheneZlExplainable-Graph-Autoformer.

Original language	English
Title of host publication	Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023
Editors	M. Arif Wani, Mihai Boicu, Moamar Sayed-Mouchaweh, Pedro Henriques Abreu, Joao Gama
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	384-391
Number of pages	8
ISBN (Electronic)	9798350345346
DOIs	https://doi.org/10.1109/ICMLA58977.2023.00060
State	Published - 2023
Event	22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023 - Jacksonville, United States Duration: Dec 15 2023 → Dec 17 2023

Publication series

Name	Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

Conference

Conference	22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023
Country/Territory	United States
City	Jacksonville
Period	12/15/23 → 12/17/23

Funding

This material is based on the work supported by the U.S. Department of Energy, Office of Science, under contract DEAC02- 06CH11357. This research used resources from the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility under contract DE-AC02- 06CH11357.

Keywords

Graph Neural Network
Long-term forecasting
Pyramid Autocorrelation attention

Access to Document

10.1109/ICMLA58977.2023.00060

Cite this

Zhong, W., Mallick, T., MacFarlane, J., Meidani, H., & Balaprakash, P. (2023). Graph Pyramid Autoformer for Long- Term Traffic Forecasting. In M. Arif Wani, M. Boicu, M. Sayed-Mouchaweh, P. H. Abreu, & J. Gama (Eds.), Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023 (pp. 384-391). (Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICMLA58977.2023.00060

Zhong, Weiheng ; Mallick, Tanwi ; MacFarlane, Jane et al. / Graph Pyramid Autoformer for Long- Term Traffic Forecasting. Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023. editor / M. Arif Wani ; Mihai Boicu ; Moamar Sayed-Mouchaweh ; Pedro Henriques Abreu ; Joao Gama. Institute of Electrical and Electronics Engineers Inc., 2023. pp. 384-391 (Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023).

@inproceedings{54590b6f72be4b99aca01fcb8ef84104,

title = "Graph Pyramid Autoformer for Long- Term Traffic Forecasting",

abstract = "Accurate traffic forecasting is vital to an intelligent transportation system. Although many deep learning models have achieved state-of-art performance for short-term traffic forecasting of up to 1 hour, long-term traffic forecasting that spans multiple hours remains a major challenge. To that end, we develop Graph Pyramid Autoformer (GPA), an attention-based spatial-temporal graph neural network that uses a novel pyramid autocorrelation attention mechanism. It enables learning from long temporal sequences on graphs and improves long-term traffic forecasting accuracy. We demonstrate the efficacy of the GPA using two benchmark traffic datasets: Los Angeles' METR-LA and the Bay Area's PEMS-BAY. Notably, our model has outperformed a range of existing state-of-the-art methods, delivering up to a 25 \% improvement in the accuracy of long-term traffic forecasts. Our code is available at: https://github.com/WeiheneZlExplainable-Graph-Autoformer.",

keywords = "Graph Neural Network, Long-term forecasting, Pyramid Autocorrelation attention",

author = "Weiheng Zhong and Tanwi Mallick and Jane MacFarlane and Hadi Meidani and Prasanna Balaprakash",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023 ; Conference date: 15-12-2023 Through 17-12-2023",

year = "2023",

doi = "10.1109/ICMLA58977.2023.00060",

language = "English",

series = "Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "384--391",

editor = "\{Arif Wani\}, M. and Mihai Boicu and Moamar Sayed-Mouchaweh and Abreu, \{Pedro Henriques\} and Joao Gama",

booktitle = "Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023",

}

Zhong, W, Mallick, T, MacFarlane, J, Meidani, H & Balaprakash, P 2023, Graph Pyramid Autoformer for Long- Term Traffic Forecasting. in M Arif Wani, M Boicu, M Sayed-Mouchaweh, PH Abreu & J Gama (eds), Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023. Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023, Institute of Electrical and Electronics Engineers Inc., pp. 384-391, 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023, Jacksonville, United States, 12/15/23. https://doi.org/10.1109/ICMLA58977.2023.00060

Graph Pyramid Autoformer for Long- Term Traffic Forecasting. / Zhong, Weiheng; Mallick, Tanwi; MacFarlane, Jane et al.
Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023. ed. / M. Arif Wani; Mihai Boicu; Moamar Sayed-Mouchaweh; Pedro Henriques Abreu; Joao Gama. Institute of Electrical and Electronics Engineers Inc., 2023. p. 384-391 (Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Graph Pyramid Autoformer for Long- Term Traffic Forecasting

AU - Zhong, Weiheng

AU - Mallick, Tanwi

AU - MacFarlane, Jane

AU - Meidani, Hadi

AU - Balaprakash, Prasanna

PY - 2023

Y1 - 2023

N2 - Accurate traffic forecasting is vital to an intelligent transportation system. Although many deep learning models have achieved state-of-art performance for short-term traffic forecasting of up to 1 hour, long-term traffic forecasting that spans multiple hours remains a major challenge. To that end, we develop Graph Pyramid Autoformer (GPA), an attention-based spatial-temporal graph neural network that uses a novel pyramid autocorrelation attention mechanism. It enables learning from long temporal sequences on graphs and improves long-term traffic forecasting accuracy. We demonstrate the efficacy of the GPA using two benchmark traffic datasets: Los Angeles' METR-LA and the Bay Area's PEMS-BAY. Notably, our model has outperformed a range of existing state-of-the-art methods, delivering up to a 25 % improvement in the accuracy of long-term traffic forecasts. Our code is available at: https://github.com/WeiheneZlExplainable-Graph-Autoformer.

AB - Accurate traffic forecasting is vital to an intelligent transportation system. Although many deep learning models have achieved state-of-art performance for short-term traffic forecasting of up to 1 hour, long-term traffic forecasting that spans multiple hours remains a major challenge. To that end, we develop Graph Pyramid Autoformer (GPA), an attention-based spatial-temporal graph neural network that uses a novel pyramid autocorrelation attention mechanism. It enables learning from long temporal sequences on graphs and improves long-term traffic forecasting accuracy. We demonstrate the efficacy of the GPA using two benchmark traffic datasets: Los Angeles' METR-LA and the Bay Area's PEMS-BAY. Notably, our model has outperformed a range of existing state-of-the-art methods, delivering up to a 25 % improvement in the accuracy of long-term traffic forecasts. Our code is available at: https://github.com/WeiheneZlExplainable-Graph-Autoformer.

KW - Graph Neural Network

KW - Long-term forecasting

KW - Pyramid Autocorrelation attention

UR - https://www.scopus.com/pages/publications/85190158197

U2 - 10.1109/ICMLA58977.2023.00060

DO - 10.1109/ICMLA58977.2023.00060

M3 - Conference contribution

AN - SCOPUS:85190158197

T3 - Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

SP - 384

EP - 391

BT - Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

A2 - Arif Wani, M.

A2 - Boicu, Mihai

A2 - Sayed-Mouchaweh, Moamar

A2 - Abreu, Pedro Henriques

A2 - Gama, Joao

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

Y2 - 15 December 2023 through 17 December 2023

ER -

Zhong W, Mallick T, MacFarlane J, Meidani H, Balaprakash P. Graph Pyramid Autoformer for Long- Term Traffic Forecasting. In Arif Wani M, Boicu M, Sayed-Mouchaweh M, Abreu PH, Gama J, editors, Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023. Institute of Electrical and Electronics Engineers Inc. 2023. p. 384-391. (Proceedings - 22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023). doi: 10.1109/ICMLA58977.2023.00060

Graph Pyramid Autoformer for Long- Term Traffic Forecasting

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this