Pre-Training General Trajectory Embeddings With Maximum Multi-View Entropy Coding

被引:2
作者
Lin, Yan [1 ,2 ]
Wan, Huaiyu [1 ,2 ]
Guo, Shengnan [1 ,2 ]
Hu, Jilin [3 ]
Jensen, Christian S.
Lin, Youfang [1 ,2 ]
机构
[1] Beijing Jiaotong Univ, Sch Comp & Informat Technoloty, Beijing Key Lab Traff Data Anal & Min, Beijing 100044, Peoples R China
[2] CAAC, Key Lab Intelligent Passenger Serv Civil Aviat, Beijing 101318, Peoples R China
[3] Aalborg Univ, Dept Comp Sci, DK-9220 Aalborg, Denmark
关键词
Trajectory; Task analysis; Roads; Semantics; Correlation; Data mining; Training; Maximum multi-view entropy; pre-training; self-supervised learning; spatio-temporal data mining; trajectory embedding; BROAD LEARNING-SYSTEM; ADAPTATION;
D O I
10.1109/TKDE.2023.3347513
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Spatio-temporal trajectories provide valuable information about movement and travel behavior, enabling various downstream tasks that in turn power real-world applications. Learning trajectory embeddings can improve task performance but may incur high computational costs and face limited training data availability. Pre-training learns generic embeddings by means of specially constructed pretext tasks that enable learning from unlabeled data. Existing pre-training methods face (i) difficulties in learning general embeddings due to biases towards certain downstream tasks incurred by the pretext tasks, (ii) limitations in capturing both travel semantics and spatio-temporal correlations, and (iii) the complexity of long, irregularly sampled trajectories. To tackle these challenges, we propose Maximum Multi-view Trajectory Entropy Coding (MMTEC) for learning general and comprehensive trajectory embeddings. We introduce a pretext task that reduces biases in pre-trained trajectory embeddings, yielding embeddings that are useful for a wide variety of downstream tasks. We also propose an attention-based discrete encoder and a NeuralCDE-based continuous encoder that extract and represent travel behavior and continuous spatio-temporal correlations from trajectories in embeddings, respectively. Extensive experiments on two real-world datasets and three downstream tasks offer insight into the design properties of our proposal and indicate that it is capable of outperforming existing trajectory embedding methods.
引用
收藏
页码:9037 / 9050
页数:14
相关论文
共 50 条
[41]   Reliable Entropy-Induced Anchor Learning for Incomplete Multi-View Subspace Clustering [J].
Shen, Qiangqiang ;
Guo, Zihou ;
Wang, Hanzhang ;
Xu, Yanhui ;
Chen, Yongyong ;
Wang, Shiqi ;
Liang, Yongsheng .
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2025, 35 (06) :5293-5306
[42]   Capsule-Guided Multi-View Attention Network for SAR Target Recognition With Small Training Set [J].
Dai, Qijun ;
Zhang, Gong ;
Xue, Biao ;
Fang, Zheng .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2023, 20
[43]   Multi-modal cross-domain self-supervised pre-training for fMRI and EEG fusion [J].
Wei, Xinxu ;
Zhao, Kanhao ;
Jiao, Yong ;
Carlisle, Nancy B. ;
Xie, Hua ;
Fonzo, Gregory A. ;
Zhang, Yu .
NEURAL NETWORKS, 2025, 184
[44]   Multi-View Collaborative Training and Self-Supervised Learning for Group Recommendation [J].
Wei, Feng ;
Chen, Shuyu .
MATHEMATICS, 2025, 13 (01)
[45]   Exploring Multi-View Pixel Contrast for General and Robust Image Forgery Localization [J].
Lou, Zijie ;
Cao, Gang ;
Guo, Kun ;
Yu, Lifang ;
Weng, Shaowei .
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2025, 20 :2329-2341
[46]   MMPT'21: International JointWorkshop on Multi-Modal Pre-Training for Multimedia Understanding [J].
Liu, Bei ;
Fu, Jianlong ;
Chen, Shizhe ;
Jin, Qin ;
Hauptmann, Alexander ;
Rui, Yong .
PROCEEDINGS OF THE 2021 INTERNATIONAL CONFERENCE ON MULTIMEDIA RETRIEVAL (ICMR '21), 2021, :694-695
[47]   mc-BEiT: Multi-choice Discretization for Image BERT Pre-training [J].
Li, Xiaotong ;
Ge, Yixiao ;
Yi, Kun ;
Hu, Zixuan ;
Shan, Ying ;
Duan, Ling-Yu .
COMPUTER VISION - ECCV 2022, PT XXX, 2022, 13690 :231-246
[48]   Gesture2Text: A Generalizable Decoder for Word-Gesture Keyboards in XR Through Trajectory Coarse Discretization and Pre-Training [J].
Shen, Junxiao ;
Khaldi, Khadija ;
Zhou, Enmin ;
Surale, Hemant Bhaskar ;
Karlson, Amy .
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, 2024, 30 (11) :7118-7128
[49]   Multi-modal Unsupervised Pre-training for Surgical Operating Room Workflow Analysis [J].
Jamal, Muhammad Abdullah ;
Mohareri, Omid .
MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION, MICCAI 2022, PT VII, 2022, 13437 :453-463
[50]   Point-LGMask: Local and Global Contexts Embedding for Point Cloud Pre-Training With Multi-Ratio Masking [J].
Tang, Yuan ;
Li, Xianzhi ;
Xu, Jinfeng ;
Yu, Qiao ;
Hu, Long ;
Hao, Yixue ;
Chen, Min .
IEEE TRANSACTIONS ON MULTIMEDIA, 2024, 26 :8360-8370