Dual-Branch Neural Networks for Predicting Shared Bikes

Huigyu Yang; Syed M. Raza; Duc Tai Le; Dongsoo S. Kim; Hyunseung Choo

doi:10.1109/IMCOM56909.2023.10035665

Dual-Branch Neural Networks for Predicting Shared Bikes

Huigyu Yang, Syed M. Raza, Duc Tai Le, Dongsoo S. Kim, Hyunseung Choo

Department of Computer Science and Engineering

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

Abstract

Bike sharing is efficient urban mobility method as a user determines the rental period upon his/her demand. A dock-based sharing system designates places (i.e., bike docks or stations) where a user rents or returns a bike. Due to regionally and temporally erratic nature of urban mobility demand, bike imbalance among stations inevitably occurs and a service operator rebalances bikes to prevent service impairment. The precise prediction of imbalance for each station is essential to optimize the rebalancing schedule. This paper presents Dual-Branch Neural Networks (DBNNs) employing Long Short-Term Memory (LSTM) to classify the imbalance level of each station along with the prediction of bike counts. To address the dual objectives, the model trains by backpropagating both classification and prediction losses to the dual output branches individually, and jointly updates the trainable parameters of LSTM layers. The results showcase that the proposed model predicts bike counts within 8% average error for next 24 hours, and achieves 94% accuracy of classification for the same time period.

Original language	English
Title of host publication	Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023
Editors	Sukhan Lee, Hyunseung Choo, Roslan Ismail
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665453486
DOIs	https://doi.org/10.1109/IMCOM56909.2023.10035665
State	Published - 2023
Event	17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023 - Seoul, Korea, Republic of Duration: 3 Jan 2023 → 5 Jan 2023

Publication series

Name	Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023

Conference

Conference	17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023
Country/Territory	Korea, Republic of
City	Seoul
Period	3/01/23 → 5/01/23

Keywords

Bike Sharing
Dual-Branch Neural Network
Time-series Prediction

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/IMCOM56909.2023.10035665

Cite this

Yang, H., Raza, S. M., Le, D. T., Kim, D. S., & Choo, H. (2023). Dual-Branch Neural Networks for Predicting Shared Bikes. In S. Lee, H. Choo, & R. Ismail (Eds.), Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023 (Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IMCOM56909.2023.10035665

Yang, Huigyu ; Raza, Syed M. ; Le, Duc Tai et al. / Dual-Branch Neural Networks for Predicting Shared Bikes. Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023. editor / Sukhan Lee ; Hyunseung Choo ; Roslan Ismail. Institute of Electrical and Electronics Engineers Inc., 2023. (Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023).

@inproceedings{6ba16c0d556545358d3e57351914f959,

title = "Dual-Branch Neural Networks for Predicting Shared Bikes",

abstract = "Bike sharing is efficient urban mobility method as a user determines the rental period upon his/her demand. A dock-based sharing system designates places (i.e., bike docks or stations) where a user rents or returns a bike. Due to regionally and temporally erratic nature of urban mobility demand, bike imbalance among stations inevitably occurs and a service operator rebalances bikes to prevent service impairment. The precise prediction of imbalance for each station is essential to optimize the rebalancing schedule. This paper presents Dual-Branch Neural Networks (DBNNs) employing Long Short-Term Memory (LSTM) to classify the imbalance level of each station along with the prediction of bike counts. To address the dual objectives, the model trains by backpropagating both classification and prediction losses to the dual output branches individually, and jointly updates the trainable parameters of LSTM layers. The results showcase that the proposed model predicts bike counts within 8\% average error for next 24 hours, and achieves 94\% accuracy of classification for the same time period.",

keywords = "Bike Sharing, Dual-Branch Neural Network, Time-series Prediction",

author = "Huigyu Yang and Raza, \{Syed M.\} and Le, \{Duc Tai\} and Kim, \{Dongsoo S.\} and Hyunseung Choo",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023 ; Conference date: 03-01-2023 Through 05-01-2023",

year = "2023",

doi = "10.1109/IMCOM56909.2023.10035665",

language = "English",

series = "Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023",

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

editor = "Sukhan Lee and Hyunseung Choo and Roslan Ismail",

booktitle = "Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023",

}

Yang, H, Raza, SM, Le, DT, Kim, DS & Choo, H 2023, Dual-Branch Neural Networks for Predicting Shared Bikes. in S Lee, H Choo & R Ismail (eds), Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023. Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023, Institute of Electrical and Electronics Engineers Inc., 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023, Seoul, Korea, Republic of, 3/01/23. https://doi.org/10.1109/IMCOM56909.2023.10035665

Dual-Branch Neural Networks for Predicting Shared Bikes. / Yang, Huigyu; Raza, Syed M.; Le, Duc Tai et al.
Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023. ed. / Sukhan Lee; Hyunseung Choo; Roslan Ismail. Institute of Electrical and Electronics Engineers Inc., 2023. (Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023).

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

TY - GEN

T1 - Dual-Branch Neural Networks for Predicting Shared Bikes

AU - Yang, Huigyu

AU - Raza, Syed M.

AU - Le, Duc Tai

AU - Kim, Dongsoo S.

AU - Choo, Hyunseung

PY - 2023

Y1 - 2023

N2 - Bike sharing is efficient urban mobility method as a user determines the rental period upon his/her demand. A dock-based sharing system designates places (i.e., bike docks or stations) where a user rents or returns a bike. Due to regionally and temporally erratic nature of urban mobility demand, bike imbalance among stations inevitably occurs and a service operator rebalances bikes to prevent service impairment. The precise prediction of imbalance for each station is essential to optimize the rebalancing schedule. This paper presents Dual-Branch Neural Networks (DBNNs) employing Long Short-Term Memory (LSTM) to classify the imbalance level of each station along with the prediction of bike counts. To address the dual objectives, the model trains by backpropagating both classification and prediction losses to the dual output branches individually, and jointly updates the trainable parameters of LSTM layers. The results showcase that the proposed model predicts bike counts within 8% average error for next 24 hours, and achieves 94% accuracy of classification for the same time period.

AB - Bike sharing is efficient urban mobility method as a user determines the rental period upon his/her demand. A dock-based sharing system designates places (i.e., bike docks or stations) where a user rents or returns a bike. Due to regionally and temporally erratic nature of urban mobility demand, bike imbalance among stations inevitably occurs and a service operator rebalances bikes to prevent service impairment. The precise prediction of imbalance for each station is essential to optimize the rebalancing schedule. This paper presents Dual-Branch Neural Networks (DBNNs) employing Long Short-Term Memory (LSTM) to classify the imbalance level of each station along with the prediction of bike counts. To address the dual objectives, the model trains by backpropagating both classification and prediction losses to the dual output branches individually, and jointly updates the trainable parameters of LSTM layers. The results showcase that the proposed model predicts bike counts within 8% average error for next 24 hours, and achieves 94% accuracy of classification for the same time period.

KW - Bike Sharing

KW - Dual-Branch Neural Network

KW - Time-series Prediction

UR - https://www.scopus.com/pages/publications/85148591845

U2 - 10.1109/IMCOM56909.2023.10035665

DO - 10.1109/IMCOM56909.2023.10035665

M3 - Conference contribution

AN - SCOPUS:85148591845

T3 - Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023

BT - Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023

A2 - Lee, Sukhan

A2 - Choo, Hyunseung

A2 - Ismail, Roslan

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023

Y2 - 3 January 2023 through 5 January 2023

ER -

Yang H, Raza SM, Le DT, Kim DS, Choo H. Dual-Branch Neural Networks for Predicting Shared Bikes. In Lee S, Choo H, Ismail R, editors, Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023. Institute of Electrical and Electronics Engineers Inc. 2023. (Proceedings of the 2023 17th International Conference on Ubiquitous Information Management and Communication, IMCOM 2023). doi: 10.1109/IMCOM56909.2023.10035665

Dual-Branch Neural Networks for Predicting Shared Bikes

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this