Predicting Future Locations of Moving Objects by Recurrent Mixture Density Network

被引:17
作者
Chen, Rui [1 ]
Chen, Mingjian [1 ]
Li, Wanli [1 ]
Guo, Naikun [1 ]
机构
[1] Informat Engn Univ, Inst Geospatial Informat, Zhengzhou 450000, Peoples R China
关键词
location prediction; long short-term memory; mixture density network; trajectory mining; BIDIRECTIONAL LSTM;
D O I
10.3390/ijgi9020116
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Accurate and timely location prediction of moving objects is crucial for intelligent transportation systems and traffic management. In recent years, ubiquitous location acquisition technologies have provided the opportunity for mining knowledge from trajectories, making location prediction and real-time decisions more feasible. Previous location prediction methods have mostly developed on the basis of shallow models whereas shallow models are not competent for some tricky challenges such as multi-time-step location coordinates prediction. Motivated by the current study status, we are dedicated to a deep-learning-based approach to predict the coordinates of several future locations of moving objects based on recent trajectory records. The method of this work consists of three successive parts: trajectory preprocessing, prediction model construction, and post-processing. In this work, a prediction model named the bidirectional recurrent mixture density network (BiRMDN) was constructed by integrating the long short-term memory (LSTM) and mixture density network (MDN) together. This model has the ability to learn long-term contextual information from recent trajectory and model real-valued location coordinates. We employed a vessel trajectory dataset for the implementation of this approach and determined the optimal model configuration after several parameter analysis experiments. Experimental results involving a performance comparison with other widely used methods demonstrate the superiority of the BiRMDN model.
引用
收藏
页数:17
相关论文
共 37 条
[1]  
[Anonymous], 2012, Improving neural networks by preventing co-adaptation of feature detectors
[2]   Using GPS to learn significant locations and predict movement across multiple users [J].
Ashbrook, Daniel ;
Starner, Thad .
PERSONAL AND UBIQUITOUS COMPUTING, 2003, 7 (05) :275-286
[3]  
Bao Jie, 2012, P 20 INT C ADV GEOGR, P199, DOI [DOI 10.1145/2424321.2424348, DOI 10.1145/2424321]
[4]   Destination Prediction by Trajectory Distribution-Based Model [J].
Besse, Philippe C. ;
Guillouet, Brendan ;
Loubes, Jean-Michel ;
Royer, Francois .
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, 2018, 19 (08) :2470-2481
[5]  
Bishop C., 1994, Technical report
[6]   Mobility Modes Awareness from Trajectories Based on Clustering and a Convolutional Neural Network [J].
Chen, Rui ;
Chen, Mingjian ;
Li, Wanli ;
Wang, Jianguang ;
Yao, Xiang .
ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION, 2019, 8 (05)
[7]   Understanding predictability and exploration in human mobility [J].
Cuttone, Andrea ;
Lehmann, Sune ;
Gonzalez, Marta C. .
EPJ DATA SCIENCE, 2018, 7
[8]   Demo: Vessel Trajectory Prediction using Sequence-to-Sequence Models over Spatial Grid [J].
Duc-Duy Nguyen ;
Van, Chan Le ;
Ali, Muhammad Intizar .
DEBS'18: PROCEEDINGS OF THE 12TH ACM INTERNATIONAL CONFERENCE ON DISTRIBUTED AND EVENT-BASED SYSTEMS, 2018, :258-261
[9]  
Endo Y., 2016, Int. J. Data Sci. Anal., V2, P107, DOI [10.1007/s41060-016-0014-1, DOI 10.1007/S41060-016-0014-1]
[10]  
Feng J., 2018, P WORLD WID WEB C LY