SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories
Pedestrian destination prediction of a user is known as an important and challenging task for LBSs (location-based services) like traffic planning and travelling recommendation. The typical method generally applies statistical model to predict the future location based on the raw trajectory. However...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2019-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8790746/ |
| _version_ | 1819298260186038272 |
|---|---|
| author | Wanlong Zhang Liting Sun Xiang Wang Zhitao Huang Baoguo Li |
| author_facet | Wanlong Zhang Liting Sun Xiang Wang Zhitao Huang Baoguo Li |
| author_sort | Wanlong Zhang |
| collection | DOAJ |
| description | Pedestrian destination prediction of a user is known as an important and challenging task for LBSs (location-based services) like traffic planning and travelling recommendation. The typical method generally applies statistical model to predict the future location based on the raw trajectory. However, while predicting, existing approaches fall short in accommodating long-range dependency and ignore the semantic information existing in the raw trajectory. In this paper, we proposed a method named semantics-enriched attentional BiGRU (SEABIG) to solve the two problems. Firstly, we designed a probabilistic model based on the GMM (Gaussian mixture model) to extract stopover points from the raw trajectories and annotate the semantic information on the stopover points. Then we proposed an attentional BiGRU-based trajectory prediction model, which can jointly learn the embeddings of the semantic trajectory. It not only takes the advantage of the BiGRU (Bidirectional Gated Recurrent Unit) for sequence modeling, but also gives more attention to meaningful positions that have strong correlations w.r.t. destination by applying attention mechanism. Finally, we annotate the most likely semantic on the predicted position with the probabilistic model. Extensive experiments on Beijing real datasets demonstrate that our proposed method has higher prediction accuracy. |
| first_indexed | 2024-12-24T05:27:04Z |
| format | Article |
| id | doaj.art-c244073696894b88a70c4f9ee715087e |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-24T05:27:04Z |
| publishDate | 2019-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-c244073696894b88a70c4f9ee715087e2022-12-21T17:13:17ZengIEEEIEEE Access2169-35362019-01-01710905410906210.1109/ACCESS.2019.29335588790746SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic TrajectoriesWanlong Zhang0https://orcid.org/0000-0002-5918-0312Liting Sun1Xiang Wang2Zhitao Huang3Baoguo Li4Department of Electronic Science and Engineering, National University of Defense Technology, Changsha, ChinaDepartment of Electronic Science and Engineering, National University of Defense Technology, Changsha, ChinaDepartment of Electronic Science and Engineering, National University of Defense Technology, Changsha, ChinaDepartment of Electronic Science and Engineering, National University of Defense Technology, Changsha, ChinaDepartment of Electronic Science and Engineering, National University of Defense Technology, Changsha, ChinaPedestrian destination prediction of a user is known as an important and challenging task for LBSs (location-based services) like traffic planning and travelling recommendation. The typical method generally applies statistical model to predict the future location based on the raw trajectory. However, while predicting, existing approaches fall short in accommodating long-range dependency and ignore the semantic information existing in the raw trajectory. In this paper, we proposed a method named semantics-enriched attentional BiGRU (SEABIG) to solve the two problems. Firstly, we designed a probabilistic model based on the GMM (Gaussian mixture model) to extract stopover points from the raw trajectories and annotate the semantic information on the stopover points. Then we proposed an attentional BiGRU-based trajectory prediction model, which can jointly learn the embeddings of the semantic trajectory. It not only takes the advantage of the BiGRU (Bidirectional Gated Recurrent Unit) for sequence modeling, but also gives more attention to meaningful positions that have strong correlations w.r.t. destination by applying attention mechanism. Finally, we annotate the most likely semantic on the predicted position with the probabilistic model. Extensive experiments on Beijing real datasets demonstrate that our proposed method has higher prediction accuracy.https://ieeexplore.ieee.org/document/8790746/Trajectory predictionsemantic trajectorydeep learning |
| spellingShingle | Wanlong Zhang Liting Sun Xiang Wang Zhitao Huang Baoguo Li SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories IEEE Access Trajectory prediction semantic trajectory deep learning |
| title | SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories |
| title_full | SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories |
| title_fullStr | SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories |
| title_full_unstemmed | SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories |
| title_short | SEABIG: A Deep Learning-Based Method for Location Prediction in Pedestrian Semantic Trajectories |
| title_sort | seabig a deep learning based method for location prediction in pedestrian semantic trajectories |
| topic | Trajectory prediction semantic trajectory deep learning |
| url | https://ieeexplore.ieee.org/document/8790746/ |
| work_keys_str_mv | AT wanlongzhang seabigadeeplearningbasedmethodforlocationpredictioninpedestriansemantictrajectories AT litingsun seabigadeeplearningbasedmethodforlocationpredictioninpedestriansemantictrajectories AT xiangwang seabigadeeplearningbasedmethodforlocationpredictioninpedestriansemantictrajectories AT zhitaohuang seabigadeeplearningbasedmethodforlocationpredictioninpedestriansemantictrajectories AT baoguoli seabigadeeplearningbasedmethodforlocationpredictioninpedestriansemantictrajectories |