Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis
The use of deep learning methods to predict traffic flow in transportation systems has become a hot research project. The existing predictive model method faces problems such as long calculation time and difficult data pre‐processing, especially for the prediction effect of high traffic area. In thi...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2020-05-01
|
| Series: | IET Intelligent Transport Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/iet-its.2019.0377 |
| _version_ | 1811281597261938688 |
|---|---|
| author | Bing Yang Yan Kang Hao Li Yachuan Zhang Yan Yang Lan Zhang |
| author_facet | Bing Yang Yan Kang Hao Li Yachuan Zhang Yan Yang Lan Zhang |
| author_sort | Bing Yang |
| collection | DOAJ |
| description | The use of deep learning methods to predict traffic flow in transportation systems has become a hot research project. The existing predictive model method faces problems such as long calculation time and difficult data pre‐processing, especially for the prediction effect of high traffic area. In this study, the authors propose a novel framework ST‐ESNet, spatio‐temporal expand‐and‐squeeze networks, that designs several effective strategies for considering the complexity, non‐linearity and uncertainty of traffic flow, and better captures traffic flow characteristics to adapt to the dynamic characteristics of traffic trajectory, traffic duration and traffic flow. Specially, we use extend‐and‐squeeze process rather than squeeze‐and‐extend process during the normal residual unit to capture farther spatial dependence among regions. Specifically, inverted residual and deformed convolution structures are utilised in the expanding process, and the convolution with stride 2 is utilised in the squeeze process. Furthermore, image feature scaling is used in each residual unit to obtain more fine‐grained surface information, which improves the ability of the model to capture dynamic spatial dependence features. Finally, they use stochastic weight averaging to obtain an integration model. In summary, they propose a new predictive model ST‐ESNet. The experimental results show that the authors’ proposed network model has better prediction performance compared with the state‐of‐the‐art model. |
| first_indexed | 2024-04-13T01:36:25Z |
| format | Article |
| id | doaj.art-6884f6d362dc481bb54c6f1d6f33a9d2 |
| institution | Directory Open Access Journal |
| issn | 1751-956X 1751-9578 |
| language | English |
| last_indexed | 2024-04-13T01:36:25Z |
| publishDate | 2020-05-01 |
| publisher | Wiley |
| record_format | Article |
| series | IET Intelligent Transport Systems |
| spelling | doaj.art-6884f6d362dc481bb54c6f1d6f33a9d22022-12-22T03:08:21ZengWileyIET Intelligent Transport Systems1751-956X1751-95782020-05-0114531332210.1049/iet-its.2019.0377Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolisBing Yang0Yan Kang1Hao Li2Yachuan Zhang3Yan Yang4Lan Zhang5Department of Software EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaDepartment of Software EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaDepartment of Network EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaDepartment of Software EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaDepartment of Software EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaDepartment of Software EngineeringSchool of Software, Yunnan UniversityKunmingPeople's Republic of ChinaThe use of deep learning methods to predict traffic flow in transportation systems has become a hot research project. The existing predictive model method faces problems such as long calculation time and difficult data pre‐processing, especially for the prediction effect of high traffic area. In this study, the authors propose a novel framework ST‐ESNet, spatio‐temporal expand‐and‐squeeze networks, that designs several effective strategies for considering the complexity, non‐linearity and uncertainty of traffic flow, and better captures traffic flow characteristics to adapt to the dynamic characteristics of traffic trajectory, traffic duration and traffic flow. Specially, we use extend‐and‐squeeze process rather than squeeze‐and‐extend process during the normal residual unit to capture farther spatial dependence among regions. Specifically, inverted residual and deformed convolution structures are utilised in the expanding process, and the convolution with stride 2 is utilised in the squeeze process. Furthermore, image feature scaling is used in each residual unit to obtain more fine‐grained surface information, which improves the ability of the model to capture dynamic spatial dependence features. Finally, they use stochastic weight averaging to obtain an integration model. In summary, they propose a new predictive model ST‐ESNet. The experimental results show that the authors’ proposed network model has better prediction performance compared with the state‐of‐the‐art model.https://doi.org/10.1049/iet-its.2019.0377crowd flow predictiondeep learning methodstraffic flow characteristicstraffic trajectorytraffic durationinverted residual convolution structures |
| spellingShingle | Bing Yang Yan Kang Hao Li Yachuan Zhang Yan Yang Lan Zhang Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis IET Intelligent Transport Systems crowd flow prediction deep learning methods traffic flow characteristics traffic trajectory traffic duration inverted residual convolution structures |
| title | Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis |
| title_full | Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis |
| title_fullStr | Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis |
| title_full_unstemmed | Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis |
| title_short | Spatio‐temporal expand‐and‐squeeze networks for crowd flow prediction in metropolis |
| title_sort | spatio temporal expand and squeeze networks for crowd flow prediction in metropolis |
| topic | crowd flow prediction deep learning methods traffic flow characteristics traffic trajectory traffic duration inverted residual convolution structures |
| url | https://doi.org/10.1049/iet-its.2019.0377 |
| work_keys_str_mv | AT bingyang spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis AT yankang spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis AT haoli spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis AT yachuanzhang spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis AT yanyang spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis AT lanzhang spatiotemporalexpandandsqueezenetworksforcrowdflowpredictioninmetropolis |