Uncertainty-Aware Temporal Graph Convolutional Network for Traffic Speed Forecasting

Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning methods have been proposed for traffic speed forecasting, the majority of them can only make point-wise prediction, which may not provide eno...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	IEEE transactions on intelligent transportation systems Jg. 25; H. 8; S. 8578 - 8590
Hauptverfasser:	Qian, Weizhu, Nielsen, Thomas Dyhre, Zhao, Yan, Larsen, Kim Guldstrand, Yu, James Jianqiao
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	IEEE 01.08.2024
Schlagworte:	Convolutional neural networks Data models Deep learning Forecasting gated recurrent unit graph convolutional network Predictive models Roads spatio-temporal model Traffic speed forecasting Uncertainty uncertainty quantification
ISSN:	1524-9050, 1558-0016
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Abstract	Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning methods have been proposed for traffic speed forecasting, the majority of them can only make point-wise prediction, which may not provide enough information for critical real-world scenarios where prediction confidence also need to be estimated, e.g., route planning for ambulances and rescue vehicles. To address this issue, we propose a novel uncertainty-aware deep learning method coined Uncertainty-Aware Temporal Graph Convolutional Network (UAT-GCN). UAT-GCN employs a Graph Convolutional Network and Gated Recurrent Unit based architecture to capture spatio-temporal dependencies. In addition, UAT-GCN consists of a specialized regressor for estimating both epistemic (model-related) and aleatoric (data-related) uncertainty. In particular, UAT-GCN utilizes Monte Carlo dropout and predictive variances to estimate epistemic and aleatoric uncertainty, respectively. In addition, we also consider the recursive dependency between predictions to further improve the forecasting performance. An extensive empirical study with real datasets offers evidence that the proposed model is capable of advancing current state-of-the-arts in terms of point-wise forecasting and quantifying prediction uncertainty with high reliability. The obtained results suggest that, compared to existing methods, the RMSE and MAE of the proposed model on the SZ-taxi dataset are reduced by 2.15% and 7.23%, respectively; the RMSE and MAE of the proposed model on the Los-loop dataset are reduced by 4.17% and 8.53%, respectively.
AbstractList	Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning methods have been proposed for traffic speed forecasting, the majority of them can only make point-wise prediction, which may not provide enough information for critical real-world scenarios where prediction confidence also need to be estimated, e.g., route planning for ambulances and rescue vehicles. To address this issue, we propose a novel uncertainty-aware deep learning method coined Uncertainty-Aware Temporal Graph Convolutional Network (UAT-GCN). UAT-GCN employs a Graph Convolutional Network and Gated Recurrent Unit based architecture to capture spatio-temporal dependencies. In addition, UAT-GCN consists of a specialized regressor for estimating both epistemic (model-related) and aleatoric (data-related) uncertainty. In particular, UAT-GCN utilizes Monte Carlo dropout and predictive variances to estimate epistemic and aleatoric uncertainty, respectively. In addition, we also consider the recursive dependency between predictions to further improve the forecasting performance. An extensive empirical study with real datasets offers evidence that the proposed model is capable of advancing current state-of-the-arts in terms of point-wise forecasting and quantifying prediction uncertainty with high reliability. The obtained results suggest that, compared to existing methods, the RMSE and MAE of the proposed model on the SZ-taxi dataset are reduced by 2.15% and 7.23%, respectively; the RMSE and MAE of the proposed model on the Los-loop dataset are reduced by 4.17% and 8.53%, respectively.
Author	Nielsen, Thomas Dyhre Larsen, Kim Guldstrand Qian, Weizhu Zhao, Yan Yu, James Jianqiao
Author_xml	– sequence: 1 givenname: Weizhu orcidid: 0000-0002-2291-4028 surname: Qian fullname: Qian, Weizhu email: wzqian@suda.edu.cn organization: School of Computer Science and Technology, Soochow University, Suzhou, China – sequence: 2 givenname: Thomas Dyhre surname: Nielsen fullname: Nielsen, Thomas Dyhre email: tdn@cs.aau.dk organization: Department of Computer Science, Aalborg University, Aalborg, Denmark – sequence: 3 givenname: Yan orcidid: 0000-0002-0242-3707 surname: Zhao fullname: Zhao, Yan email: yanz@cs.aau.dk organization: Department of Computer Science, Aalborg University, Aalborg, Denmark – sequence: 4 givenname: Kim Guldstrand orcidid: 0000-0002-5953-3384 surname: Larsen fullname: Larsen, Kim Guldstrand email: kgl@cs.aau.dk organization: Department of Computer Science, Aalborg University, Aalborg, Denmark – sequence: 5 givenname: James Jianqiao orcidid: 0000-0002-6392-6711 surname: Yu fullname: Yu, James Jianqiao email: jqyu@ieee.org organization: Department of Computer Science, University of York, York, U.K
BookMark	eNp9kMtKAzEUhoNUsK0-gOBiXmBq7plZlmIvUHTRcT2cpolGp8mQiZa-vR3ahbhwdQ4_fP_hfCM08MEbhO4JnhCCy8dqVW0mFFM-YUwKRckVGhIhihxjIgf9TnleYoFv0KjrPk4pF4QMUfXqtYkJnE_HfHqAaLLK7NsQockWEdr3bBb8d2i-kgv-lD2bdAjxM7MhZlUEa53ONq0xu2weotHQJeffbtG1haYzd5c5RtX8qZot8_XLYjWbrnNNpUy55QXIAqgpuOIYpAKBS1kKIMVOW2UJs7KEraZsp5QEBUC42m4pBiYlU2yMyLlWx9B10di6jW4P8VgTXPdW6t5K3VupL1ZOjPrDaJegfy5FcM2_5MOZdMaYX5e4KDCm7AeygnK4
CODEN	ITISFG
CitedBy_id	crossref_primary_10_1016_j_aap_2025_108238 crossref_primary_10_1111_mice_70078 crossref_primary_10_1007_s12145_025_01964_0 crossref_primary_10_3390_math12213338 crossref_primary_10_1080_01441647_2024_2408840 crossref_primary_10_1109_TITS_2025_3551830 crossref_primary_10_1016_j_eswa_2025_129223
Cites_doi	10.1109/ACCESS.2021.3062114 10.1080/01621459.2017.1307116 10.1109/TKDE.2020.3025580 10.1257/jep.15.4.143 10.24963/ijcai.2019/264 10.1609/aaai.v33i01.3301922 10.5555/3045390.3045502 10.1109/TITS.2019.2935152 10.48550/ARXIV.1706.03762 10.1609/aaai.v34i01.5477 10.48550/arXiv.1609.03499 10.1609/aaai.v35i17.17761 10.1162/neco.1989.1.2.270 10.1145/3394486.3403118 10.24963/ijcai.2018/505 10.1109/TITS.2020.3008612 10.3115/v1/W14-4012 10.1109/TITS.2019.2929020 10.1016/j.inffus.2021.05.008 10.1109/ICNN.1994.374138 10.48550/ARXIV.1609.02907 10.1007/978-3-540-28650-9_4 10.1109/TKDE.2021.3117986 10.1145/3447548.3467325 10.1162/neco.1997.9.8.1735
ContentType	Journal Article
DBID	97E RIA RIE AAYXX CITATION
DOI	10.1109/TITS.2024.3365721
DatabaseName	IEEE Xplore (IEEE) IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Electronic Library (IEL) CrossRef
DatabaseTitle	CrossRef
DatabaseTitleList
Database_xml	– sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher
DeliveryMethod	fulltext_linktorsrc
Discipline	Engineering
EISSN	1558-0016
EndPage	8590
ExternalDocumentID	10_1109_TITS_2024_3365721 10458002
Genre	orig-research
GroupedDBID	-~X 0R~ 29I 4.4 5GY 5VS 6IK 97E AAJGR AARMG AASAJ AAWTH ABAZT ABQJQ ABVLG ACGFO ACGFS ACIWK ACNCT AENEX AETIX AGQYO AGSQL AHBIQ AIBXA AKJIK AKQYR ALMA_UNASSIGNED_HOLDINGS ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ CS3 DU5 EBS EJD HZ~ H~9 IFIPE IPLJI JAVBF LAI M43 O9- OCL P2P PQQKQ RIA RIE RNS ZY4 AAYXX CITATION
ID	FETCH-LOGICAL-c266t-f48a68a2e84740a67a509695a18dcf7f13f69abc23d776a7aa147bb20a366373
IEDL.DBID	RIE
ISICitedReferencesCount	9
ISICitedReferencesURI	http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001181536200001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN	1524-9050
IngestDate	Tue Nov 18 22:35:39 EST 2025 Sat Nov 29 06:35:06 EST 2025 Wed Aug 27 02:33:47 EDT 2025
IsPeerReviewed	true
IsScholarly	true
Issue	8
Language	English
License	https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c266t-f48a68a2e84740a67a509695a18dcf7f13f69abc23d776a7aa147bb20a366373
ORCID	0000-0002-0242-3707 0000-0002-5953-3384 0000-0002-6392-6711 0000-0002-2291-4028
PageCount	13
ParticipantIDs	crossref_citationtrail_10_1109_TITS_2024_3365721 crossref_primary_10_1109_TITS_2024_3365721 ieee_primary_10458002
PublicationCentury	2000
PublicationDate	2024-08-01
PublicationDateYYYYMMDD	2024-08-01
PublicationDate_xml	– month: 08 year: 2024 text: 2024-08-01 day: 01
PublicationDecade	2020
PublicationTitle	IEEE transactions on intelligent transportation systems
PublicationTitleAbbrev	TITS
PublicationYear	2024
Publisher	IEEE
Publisher_xml	– name: IEEE
References	ref35 ref12 ref15 ref14 Stankeviciute (ref25); 34 ref36 Damianou (ref29) ref11 ref10 Garipov (ref31); 31 ref1 Lakshminarayanan (ref30); 30 Jiang (ref2) 2021 ref17 ref39 ref16 ref19 ref18 Srivastava (ref38) 2014; 15 Izmailov (ref27) Jospin (ref26) 2020 Izmailov (ref33); 2 Blundell (ref7) Pearce (ref42) Kendall (ref6) 2017 Wilson (ref34); 33 Fort (ref32) 2019 ref24 ref23 ref20 ref41 ref22 ref21 ref8 ref9 ref4 ref3 ref5 Welling (ref28) ref40 Li (ref13) Snelson (ref37)
References_xml	– start-page: 524 volume-title: Proc. Artif. Intell. Statist. ident: ref37 article-title: Local and global sparse Gaussian process approximations – ident: ref41 doi: 10.1109/ACCESS.2021.3062114 – ident: ref24 doi: 10.1080/01621459.2017.1307116 – volume: 2 start-page: 876 volume-title: Proc. 34th Conf. Uncertainty Artif. Intell. ident: ref33 article-title: Averaging weights leads to wider optima and better generalization – volume: 33 start-page: 4697 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref34 article-title: Bayesian deep learning and a probabilistic perspective of generalization – ident: ref4 doi: 10.1109/TKDE.2020.3025580 – ident: ref23 doi: 10.1257/jep.15.4.143 – year: 2020 ident: ref26 article-title: Hands-on Bayesian neural networks—A tutorial for deep learning users publication-title: arXiv:2007.06823 – ident: ref15 doi: 10.24963/ijcai.2019/264 – ident: ref18 doi: 10.1609/aaai.v33i01.3301922 – ident: ref9 doi: 10.5555/3045390.3045502 – ident: ref14 doi: 10.1109/TITS.2019.2935152 – ident: ref17 doi: 10.48550/ARXIV.1706.03762 – ident: ref19 doi: 10.1609/aaai.v34i01.5477 – start-page: 4629 volume-title: Proc. Int. Conf. Mach. Learn. ident: ref27 article-title: What are Bayesian neural network posteriors really like? – year: 2017 ident: ref6 article-title: What uncertainties do we need in Bayesian deep learning for computer vision? publication-title: arXiv:1703.04977 – ident: ref16 doi: 10.48550/arXiv.1609.03499 – ident: ref21 doi: 10.1609/aaai.v35i17.17761 – ident: ref39 doi: 10.1162/neco.1989.1.2.270 – volume: 15 start-page: 1929 year: 2014 ident: ref38 article-title: Dropout: A simple way to prevent neural networks from overfitting publication-title: J. Mach. Learn. Res. – ident: ref40 doi: 10.1145/3394486.3403118 – ident: ref20 doi: 10.24963/ijcai.2018/505 – year: 2021 ident: ref2 article-title: Graph neural network for traffic forecasting: A survey publication-title: arXiv:2101.11174 – year: 2019 ident: ref32 article-title: Deep ensembles: A loss landscape perspective publication-title: arXiv:1912.02757 – ident: ref3 doi: 10.1109/TITS.2020.3008612 – ident: ref12 doi: 10.3115/v1/W14-4012 – ident: ref1 doi: 10.1109/TITS.2019.2929020 – start-page: 4075 volume-title: Proc. Int. Conf. Mach. Learn. ident: ref42 article-title: High-quality prediction intervals for deep learning: A distribution-free, ensembled approach – ident: ref5 doi: 10.1016/j.inffus.2021.05.008 – ident: ref22 doi: 10.1109/ICNN.1994.374138 – volume: 31 start-page: 8803 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref31 article-title: Loss surfaces, mode connectivity, and fast ensembling of DNNs – volume: 34 start-page: 6216 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref25 article-title: Conformal time-series forecasting – start-page: 207 volume-title: Proc. Artif. Intell. Stat. ident: ref29 article-title: Deep Gaussian processes – ident: ref11 doi: 10.48550/ARXIV.1609.02907 – ident: ref8 doi: 10.1007/978-3-540-28650-9_4 – start-page: 681 volume-title: Proc. 28th Int. Conf. Mach. Learn. (ICML) ident: ref28 article-title: Bayesian learning via stochastic gradient Langevin dynamics – ident: ref35 doi: 10.1109/TKDE.2021.3117986 – start-page: 1 volume-title: Proc. Int. Conf. Learn. Represent. ident: ref13 article-title: Diffusion convolutional recurrent neural network: Data-driven traffic forecasting – volume: 30 start-page: 6402 volume-title: Proc. Adv. Neural Inf. Process. Syst. ident: ref30 article-title: Simple and scalable predictive uncertainty estimation using deep ensembles – ident: ref10 doi: 10.1145/3447548.3467325 – start-page: 1613 volume-title: Proc. Int. Conf. Mach. Learn. ident: ref7 article-title: Weight uncertainty in neural network – ident: ref36 doi: 10.1162/neco.1997.9.8.1735
SSID	ssj0014511
Score	2.472464
Snippet	Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning...
SourceID	crossref ieee
SourceType	Enrichment Source Index Database Publisher
StartPage	8578
SubjectTerms	Convolutional neural networks Data models Deep learning Forecasting gated recurrent unit graph convolutional network Predictive models Roads spatio-temporal model Traffic speed forecasting Uncertainty uncertainty quantification
Title	Uncertainty-Aware Temporal Graph Convolutional Network for Traffic Speed Forecasting
URI	https://ieeexplore.ieee.org/document/10458002
Volume	25
WOSCitedRecordID	wos001181536200001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText	1
inHoldings	1
isFullTextHit
isPrint
journalDatabaseRights	– providerCode: PRVIEE databaseName: IEEE Electronic Library (IEL) customDbUrl: eissn: 1558-0016 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0014511 issn: 1524-9050 databaseCode: RIE dateStart: 20000101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3dS8MwEA86fNAHPyfOL_Lgk9DZNGnSPo7hF8gQVmVv5ZqmIIxubN3E_95L2o35oOBLCSEh5ZeEu8vd_Y6QGy2j2Bi0TnLBtCfCWOCVAoYfZdCegEI7Jqb3FzUYRKNR_Nokq7tcGGOMCz4zXdt0vvx8ohf2qQxvuAgjRx25rZSsk7XWLgNLtOXIUQNczg9XLkzmx3fJczJEUzAQXc5lqAL2QwhtVFVxQuXh4J-_c0j2G-2R9urtPiJbpjwmexucgickecNtdG7-6svrfcLM0KSmnxrTR8tOTfuTctkcOOwb1HHgFJVXioLLMkrQ4RRlGrVVOzXMbVx0myQP90n_yWtKJ3gaJW7lFSICGUGAeCtEXCqwNC9xCCzKdaEKxgsZQ6YDniOSoACYUFkW-MBRBVH8lLTKSWnOCNWiCLkUfsa5FjmXoLngecR4HoBiBe8QfwVlqhtacVvdYpw688KPU4t-atFPG_Q75HY9ZVpzavw1uG2R3xhYg37-S_8F2bXT6xi9S9KqZgtzRXb0svqYz67dkfkGp6i9pA
linkProvider	IEEE
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PS8MwFA4yBfXgz4nzZw6ehM6mSZP2OIZzw1mEVdmtZGkKwujG1k38731JuzEPCl5KCSktXxLee33vfR9Cd4oHodYQnaSMKIf5IYMjJQlchIZ4QmbKMjG990UUBcNh-Fo1q9teGK21LT7TTXNrc_npRC3MrzI44cwPLHXktpHOqtq11kkDQ7Vl6VE9eKHrr5KYxA0f4l48gGDQY01KuS888sMMbeiqWLPSOfznBx2hg8p_xK1ywY_Rls5P0P4Gq-Apit9gIW2iv_hyWp9ypnFcElCN8ZPhp8btSb6sthyMRWUlOAb3FYPpMpwSeDAFq4aNbqeSc1MZXUdx5zFud51KPMFRYHMLJ2OB5IH0AHEBmHMhDdFL6EsSpCoTGaEZD-VIeTQVgkshJWFiNPJcScEJEfQM1fJJrs8RVizzKWfuiFLFUsqlooymAaGpJwXJaAO5KygTVRGLG32LcWIDDDdMDPqJQT-p0G-g-_Uj05JV46_JdYP8xsQS9Itfxm_Rbjd-6Sf9XvR8ifa8VdEeuUK1YrbQ12hHLYuP-ezGbp9vtZrA4g
openUrl	ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Uncertainty-Aware+Temporal+Graph+Convolutional+Network+for+Traffic+Speed+Forecasting&rft.jtitle=IEEE+transactions+on+intelligent+transportation+systems&rft.au=Qian%2C+Weizhu&rft.au=Nielsen%2C+Thomas+Dyhre&rft.au=Zhao%2C+Yan&rft.au=Larsen%2C+Kim+Guldstrand&rft.date=2024-08-01&rft.pub=IEEE&rft.issn=1524-9050&rft.volume=25&rft.issue=8&rft.spage=8578&rft.epage=8590&rft_id=info:doi/10.1109%2FTITS.2024.3365721&rft.externalDocID=10458002
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1524-9050&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1524-9050&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1524-9050&client=summon