Efficient Online Hyperparameter Learning for Traffic Flow Prediction

Hongyuan Zhan, Gabriel Gomes, Xiaoye S. Li, Kamesh Madduri, Kesheng Wu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Compute efficiency is an important consideration for traffic flow prediction models. Machine learning algorithms adjust model parameters automatically based on the data, but often require users to set additional parameters, known as hyperparameters. Hyperparameters can significantly impact prediction accuracy. Traffic measurements, typically collected online by sensors, are serially correlated. Moreover, the data distribution may change gradually. A typical adaptation strategy is periodically re-tuning the model hyperparameters, at the cost of computational burden. In this work, we present an efficient and principled online hyperparameter learning algorithm for kernel-based traffic prediction models. In tests with real traffic measurement data, our approach requires as little as one-seventh of the computation time of other tuning methods, while achieving better or similar prediction accuracy.

Original languageEnglish (US)
Title of host publication2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages164-169
Number of pages6
ISBN (Electronic)9781728103235
DOIs
StatePublished - Dec 7 2018
Event21st IEEE International Conference on Intelligent Transportation Systems, ITSC 2018 - Maui, United States
Duration: Nov 4 2018Nov 7 2018

Publication series

NameIEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC
Volume2018-November

Other

Other21st IEEE International Conference on Intelligent Transportation Systems, ITSC 2018
CountryUnited States
CityMaui
Period11/4/1811/7/18

Fingerprint

Learning algorithms
Tuning
Learning systems
Sensors

All Science Journal Classification (ASJC) codes

  • Automotive Engineering
  • Mechanical Engineering
  • Computer Science Applications

Cite this

Zhan, H., Gomes, G., Li, X. S., Madduri, K., & Wu, K. (2018). Efficient Online Hyperparameter Learning for Traffic Flow Prediction. In 2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018 (pp. 164-169). [8569972] (IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC; Vol. 2018-November). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ITSC.2018.8569972
Zhan, Hongyuan ; Gomes, Gabriel ; Li, Xiaoye S. ; Madduri, Kamesh ; Wu, Kesheng. / Efficient Online Hyperparameter Learning for Traffic Flow Prediction. 2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 164-169 (IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC).
@inproceedings{e135a594d27f481581af20d953923634,
title = "Efficient Online Hyperparameter Learning for Traffic Flow Prediction",
abstract = "Compute efficiency is an important consideration for traffic flow prediction models. Machine learning algorithms adjust model parameters automatically based on the data, but often require users to set additional parameters, known as hyperparameters. Hyperparameters can significantly impact prediction accuracy. Traffic measurements, typically collected online by sensors, are serially correlated. Moreover, the data distribution may change gradually. A typical adaptation strategy is periodically re-tuning the model hyperparameters, at the cost of computational burden. In this work, we present an efficient and principled online hyperparameter learning algorithm for kernel-based traffic prediction models. In tests with real traffic measurement data, our approach requires as little as one-seventh of the computation time of other tuning methods, while achieving better or similar prediction accuracy.",
author = "Hongyuan Zhan and Gabriel Gomes and Li, {Xiaoye S.} and Kamesh Madduri and Kesheng Wu",
year = "2018",
month = "12",
day = "7",
doi = "10.1109/ITSC.2018.8569972",
language = "English (US)",
series = "IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "164--169",
booktitle = "2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018",
address = "United States",

}

Zhan, H, Gomes, G, Li, XS, Madduri, K & Wu, K 2018, Efficient Online Hyperparameter Learning for Traffic Flow Prediction. in 2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018., 8569972, IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC, vol. 2018-November, Institute of Electrical and Electronics Engineers Inc., pp. 164-169, 21st IEEE International Conference on Intelligent Transportation Systems, ITSC 2018, Maui, United States, 11/4/18. https://doi.org/10.1109/ITSC.2018.8569972

Efficient Online Hyperparameter Learning for Traffic Flow Prediction. / Zhan, Hongyuan; Gomes, Gabriel; Li, Xiaoye S.; Madduri, Kamesh; Wu, Kesheng.

2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 164-169 8569972 (IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC; Vol. 2018-November).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Efficient Online Hyperparameter Learning for Traffic Flow Prediction

AU - Zhan, Hongyuan

AU - Gomes, Gabriel

AU - Li, Xiaoye S.

AU - Madduri, Kamesh

AU - Wu, Kesheng

PY - 2018/12/7

Y1 - 2018/12/7

N2 - Compute efficiency is an important consideration for traffic flow prediction models. Machine learning algorithms adjust model parameters automatically based on the data, but often require users to set additional parameters, known as hyperparameters. Hyperparameters can significantly impact prediction accuracy. Traffic measurements, typically collected online by sensors, are serially correlated. Moreover, the data distribution may change gradually. A typical adaptation strategy is periodically re-tuning the model hyperparameters, at the cost of computational burden. In this work, we present an efficient and principled online hyperparameter learning algorithm for kernel-based traffic prediction models. In tests with real traffic measurement data, our approach requires as little as one-seventh of the computation time of other tuning methods, while achieving better or similar prediction accuracy.

AB - Compute efficiency is an important consideration for traffic flow prediction models. Machine learning algorithms adjust model parameters automatically based on the data, but often require users to set additional parameters, known as hyperparameters. Hyperparameters can significantly impact prediction accuracy. Traffic measurements, typically collected online by sensors, are serially correlated. Moreover, the data distribution may change gradually. A typical adaptation strategy is periodically re-tuning the model hyperparameters, at the cost of computational burden. In this work, we present an efficient and principled online hyperparameter learning algorithm for kernel-based traffic prediction models. In tests with real traffic measurement data, our approach requires as little as one-seventh of the computation time of other tuning methods, while achieving better or similar prediction accuracy.

UR - http://www.scopus.com/inward/record.url?scp=85060483503&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060483503&partnerID=8YFLogxK

U2 - 10.1109/ITSC.2018.8569972

DO - 10.1109/ITSC.2018.8569972

M3 - Conference contribution

AN - SCOPUS:85060483503

T3 - IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC

SP - 164

EP - 169

BT - 2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Zhan H, Gomes G, Li XS, Madduri K, Wu K. Efficient Online Hyperparameter Learning for Traffic Flow Prediction. In 2018 IEEE Intelligent Transportation Systems Conference, ITSC 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 164-169. 8569972. (IEEE Conference on Intelligent Transportation Systems, Proceedings, ITSC). https://doi.org/10.1109/ITSC.2018.8569972