From Moves to Paths: A Hierarchical Framework for Trajectory Representation Learning

Chundong Wang, Xiangtian Zheng, Qingbo Hao, Yongxin Zhao, Yixuan Song, Jia Li
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:7121-7136, 2026.

Abstract

Trajectory representation learning (TRL) seeks to convert trajectory data into low-dimensional embeddings for various downstream tasks. Existing methods are limited to a single perspective: GPS-based approaches capture dynamic details but lack semantic context, while route-based methods preserve structure but lose fine-grained motion patterns. To address these limitations, we propose a trajectory representation model with multi-perspective fusion, MPH, where \underline{M} denotes the Motion modality (GPS), \underline{P} denotes the planned path modality (Route), and\underline{H} signifies the Hierarchical encoding and fusion strategy. MPH first map-matches raw GPS trajectories to route sequences and designs dedicated encoders to extract dynamic behavioral patterns and static semantic features respectively. Based on this, MPH leverages a cross-attention mechanism for modality interaction and fusion, producing fused route representations that are hierarchically aggregated into trajectory-level embeddings. Furthermore, we introduce two self-supervised tasks to train the model. Contrastive learning is employed to align road segment representations across the two perspectives, while the dual-mask prediction task strengthens contextual modeling by jointly reconstructing masked road segment identities and their corresponding temporal information. Experiments on two real-world datasets show that MPH outperforms all baselines across different tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-wang26e, title = {From Moves to Paths: A Hierarchical Framework for Trajectory Representation Learning}, author = {Wang, Chundong and Zheng, Xiangtian and Hao, Qingbo and Zhao, Yongxin and Song, Yixuan and Li, Jia}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {7121--7136}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/wang26e/wang26e.pdf}, url = {https://proceedings.mlr.press/v337/wang26e.html}, abstract = {Trajectory representation learning (TRL) seeks to convert trajectory data into low-dimensional embeddings for various downstream tasks. Existing methods are limited to a single perspective: GPS-based approaches capture dynamic details but lack semantic context, while route-based methods preserve structure but lose fine-grained motion patterns. To address these limitations, we propose a trajectory representation model with multi-perspective fusion, MPH, where \underline{M} denotes the Motion modality (GPS), \underline{P} denotes the planned path modality (Route), and\underline{H} signifies the Hierarchical encoding and fusion strategy. MPH first map-matches raw GPS trajectories to route sequences and designs dedicated encoders to extract dynamic behavioral patterns and static semantic features respectively. Based on this, MPH leverages a cross-attention mechanism for modality interaction and fusion, producing fused route representations that are hierarchically aggregated into trajectory-level embeddings. Furthermore, we introduce two self-supervised tasks to train the model. Contrastive learning is employed to align road segment representations across the two perspectives, while the dual-mask prediction task strengthens contextual modeling by jointly reconstructing masked road segment identities and their corresponding temporal information. Experiments on two real-world datasets show that MPH outperforms all baselines across different tasks.} }
Endnote
%0 Conference Paper %T From Moves to Paths: A Hierarchical Framework for Trajectory Representation Learning %A Chundong Wang %A Xiangtian Zheng %A Qingbo Hao %A Yongxin Zhao %A Yixuan Song %A Jia Li %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-wang26e %I PMLR %P 7121--7136 %U https://proceedings.mlr.press/v337/wang26e.html %V 337 %X Trajectory representation learning (TRL) seeks to convert trajectory data into low-dimensional embeddings for various downstream tasks. Existing methods are limited to a single perspective: GPS-based approaches capture dynamic details but lack semantic context, while route-based methods preserve structure but lose fine-grained motion patterns. To address these limitations, we propose a trajectory representation model with multi-perspective fusion, MPH, where \underline{M} denotes the Motion modality (GPS), \underline{P} denotes the planned path modality (Route), and\underline{H} signifies the Hierarchical encoding and fusion strategy. MPH first map-matches raw GPS trajectories to route sequences and designs dedicated encoders to extract dynamic behavioral patterns and static semantic features respectively. Based on this, MPH leverages a cross-attention mechanism for modality interaction and fusion, producing fused route representations that are hierarchically aggregated into trajectory-level embeddings. Furthermore, we introduce two self-supervised tasks to train the model. Contrastive learning is employed to align road segment representations across the two perspectives, while the dual-mask prediction task strengthens contextual modeling by jointly reconstructing masked road segment identities and their corresponding temporal information. Experiments on two real-world datasets show that MPH outperforms all baselines across different tasks.
APA
Wang, C., Zheng, X., Hao, Q., Zhao, Y., Song, Y. & Li, J.. (2026). From Moves to Paths: A Hierarchical Framework for Trajectory Representation Learning. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:7121-7136 Available from https://proceedings.mlr.press/v337/wang26e.html.

Related Material