MF-CNN: Traffic Flow Prediction Using Convolutional Neural Network and Multi-Features Fusion

Di YANG; Songjiang LI; Zhou PENG; Peng WANG; Junhui WANG; Huamin YANG

doi:10.1587/transinf.2018EDP7330

MF-CNN: Traffic Flow Prediction Using Convolutional Neural Network and Multi-Features Fusion

Di YANG, Songjiang LI, Zhou PENG, Peng WANG, Junhui WANG, Huamin YANG

Full Text Views

0

Cite this

Summary :

Accurate traffic flow prediction is the precondition for many applications in Intelligent Transportation Systems, such as traffic control and route guidance. Traditional data driven traffic flow prediction models tend to ignore traffic self-features (e.g., periodicities), and commonly suffer from the shifts brought by various complex factors (e.g., weather and holidays). These would reduce the precision and robustness of the prediction models. To tackle this problem, in this paper, we propose a CNN-based multi-feature predictive model (MF-CNN) that collectively predicts network-scale traffic flow with multiple spatiotemporal features and external factors (weather and holidays). Specifically, we classify traffic self-features into temporal continuity as short-term feature, daily periodicity and weekly periodicity as long-term features, then map them to three two-dimensional spaces, which each one is composed of time and space, represented by two-dimensional matrices. The high-level spatiotemporal features learned by CNNs from the matrices with different time lags are further fused with external factors by a logistic regression layer to derive the final prediction. Experimental results indicate that the MF-CNN model considering multi-features improves the predictive performance compared to five baseline models, and achieves the trade-off between accuracy and efficiency.

Publication: IEICE TRANSACTIONS on Information Vol.E102-D No.8 pp.1526-1536

Publication Date: 2019/08/01

Publicized: 2019/05/20

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2018EDP7330

Type of Manuscript: PAPER

Category: Artificial Intelligence, Data Mining

Authors

Di YANG
  Changchun University of Science and Technology
Songjiang LI
  Changchun University of Science and Technology
Zhou PENG
  Changchun University of Science and Technology
Peng WANG
  Changchun University of Science and Technology
Junhui WANG
  Changchun
Huamin YANG
  Changchun University of Science and Technology

Keyword

Intelligent Transportation Systems, traffic flow prediction, convolutional neural networks, multiple spatiotemporal features, external factors

Cite this

Copy

Di YANG, Songjiang LI, Zhou PENG, Peng WANG, Junhui WANG, Huamin YANG, "MF-CNN: Traffic Flow Prediction Using Convolutional Neural Network and Multi-Features Fusion" in IEICE TRANSACTIONS on Information, vol. E102-D, no. 8, pp. 1526-1536, August 2019, doi: 10.1587/transinf.2018EDP7330.
Abstract: Accurate traffic flow prediction is the precondition for many applications in Intelligent Transportation Systems, such as traffic control and route guidance. Traditional data driven traffic flow prediction models tend to ignore traffic self-features (e.g., periodicities), and commonly suffer from the shifts brought by various complex factors (e.g., weather and holidays). These would reduce the precision and robustness of the prediction models. To tackle this problem, in this paper, we propose a CNN-based multi-feature predictive model (MF-CNN) that collectively predicts network-scale traffic flow with multiple spatiotemporal features and external factors (weather and holidays). Specifically, we classify traffic self-features into temporal continuity as short-term feature, daily periodicity and weekly periodicity as long-term features, then map them to three two-dimensional spaces, which each one is composed of time and space, represented by two-dimensional matrices. The high-level spatiotemporal features learned by CNNs from the matrices with different time lags are further fused with external factors by a logistic regression layer to derive the final prediction. Experimental results indicate that the MF-CNN model considering multi-features improves the predictive performance compared to five baseline models, and achieves the trade-off between accuracy and efficiency.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018EDP7330/_p

Copy

@ARTICLE{e102-d_8_1526,
author={Di YANG, Songjiang LI, Zhou PENG, Peng WANG, Junhui WANG, Huamin YANG, },
journal={IEICE TRANSACTIONS on Information},
title={MF-CNN: Traffic Flow Prediction Using Convolutional Neural Network and Multi-Features Fusion},
year={2019},
volume={E102-D},
number={8},
pages={1526-1536},
abstract={Accurate traffic flow prediction is the precondition for many applications in Intelligent Transportation Systems, such as traffic control and route guidance. Traditional data driven traffic flow prediction models tend to ignore traffic self-features (e.g., periodicities), and commonly suffer from the shifts brought by various complex factors (e.g., weather and holidays). These would reduce the precision and robustness of the prediction models. To tackle this problem, in this paper, we propose a CNN-based multi-feature predictive model (MF-CNN) that collectively predicts network-scale traffic flow with multiple spatiotemporal features and external factors (weather and holidays). Specifically, we classify traffic self-features into temporal continuity as short-term feature, daily periodicity and weekly periodicity as long-term features, then map them to three two-dimensional spaces, which each one is composed of time and space, represented by two-dimensional matrices. The high-level spatiotemporal features learned by CNNs from the matrices with different time lags are further fused with external factors by a logistic regression layer to derive the final prediction. Experimental results indicate that the MF-CNN model considering multi-features improves the predictive performance compared to five baseline models, and achieves the trade-off between accuracy and efficiency.},
keywords={},
doi={10.1587/transinf.2018EDP7330},
ISSN={1745-1361},
month={August},}

Copy

TY - JOUR
TI - MF-CNN: Traffic Flow Prediction Using Convolutional Neural Network and Multi-Features Fusion
T2 - IEICE TRANSACTIONS on Information
SP - 1526
EP - 1536
AU - Di YANG
AU - Songjiang LI
AU - Zhou PENG
AU - Peng WANG
AU - Junhui WANG
AU - Huamin YANG
PY - 2019
DO - 10.1587/transinf.2018EDP7330
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 8
JA - IEICE TRANSACTIONS on Information
Y1 - August 2019
AB - Accurate traffic flow prediction is the precondition for many applications in Intelligent Transportation Systems, such as traffic control and route guidance. Traditional data driven traffic flow prediction models tend to ignore traffic self-features (e.g., periodicities), and commonly suffer from the shifts brought by various complex factors (e.g., weather and holidays). These would reduce the precision and robustness of the prediction models. To tackle this problem, in this paper, we propose a CNN-based multi-feature predictive model (MF-CNN) that collectively predicts network-scale traffic flow with multiple spatiotemporal features and external factors (weather and holidays). Specifically, we classify traffic self-features into temporal continuity as short-term feature, daily periodicity and weekly periodicity as long-term features, then map them to three two-dimensional spaces, which each one is composed of time and space, represented by two-dimensional matrices. The high-level spatiotemporal features learned by CNNs from the matrices with different time lags are further fused with external factors by a logistic regression layer to derive the final prediction. Experimental results indicate that the MF-CNN model considering multi-features improves the predictive performance compared to five baseline models, and achieves the trade-off between accuracy and efficiency.
ER -