Pith. sign in

REVIEW 4 major objections 5 minor 41 references

TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read TransferTraj claims that a single pre-trained trajectory encoder transfers across cities and across trajectory prediction, recovery, and travel-time tasks without retraining.

desk verdict A genuinely useful task-transferable masking scheme with solid few-shot results, but the zero-shot TP headline rests on an unfair baseline comparison and an internal hyperparameter conflict. read the letter →

arxiv 2505.12672 v1 pith:4WQCQHXN submitted 2025-05-19 cs.LG

Tonglong Wei , Yan Lin , Zeyu Zhou , Haomin Wen , Jilin Hu , Shengnan Guo , Youfang Lin , Gao Cong
show 1 more author
Huaiyu Wan
This is my paper · ORCID
classification cs.LG
keywords vehicletrajectorylearningregiontransferabilitytaskpredictionrecoverytraveltimeestimationmixture-of-expertsrotarypositionembedding
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

TransferTraj claims that a single vehicle-trajectory model can be pre-trained once and then applied to new regions and new prediction tasks without retraining. The model represents each GPS point through four modalities—location, time, nearby points of interest, and nearby road segments—and uses that context to make movement patterns transferable across cities. A masking-and-recovery pre-training scheme rewrites trajectory prediction, trajectory recovery, and origin-destination travel-time estimation as the same input-output operation, so one set of weights serves all three. If the claim holds, regions with little trajectory data and tasks with scarce labels can inherit a capable model instead of training from scratch.

What carries the argument

The load-bearing object is the Region-Transferable Trajectory Encoder (RTTE), built from TRIE and SC-MoE. TRIE places a learnable rotation matrix on attention queries and keys so that the dot product between two trajectory points encodes their relative spatial displacement rather than their absolute city coordinates. SC-MoE uses noisy top-k gating to assign each trajectory point to movement-pattern experts based on the density of nearby points of interest and road segments, letting similar spatial contexts share experts across regions. The task-transferable input-output scheme wraps these in a single masking-and-recovery interface, which is what allows one pre-trained model to serve multiple tasks.

What would settle it

Cluster the POI and road text embeddings used in Section 4.1.1 by city; if city identity explains more variance than semantic category, the region-transfer mechanism loses its grounding, and zero-shot gains should shrink when a text encoder without knowledge of the target city is substituted.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that region transfer and task transfer can be built into a single encoder plus a single pre-training objective rather than added as separate modules. TransferTraj's RTTE encoder represents each trajectory point by its offset from the trajectory start, its temporal features, and text-derived embeddings of nearby points of interest and road segments; a rotary relative-position mechanism (TRIE) makes attention depend on relative displacement instead of absolute coordinates, and a spatial-context mixture-of-experts (SC-MoE) routes movement patterns to specialists selected by local context density. The task-transferable scheme replaces task-specific heads with one operation: mask either a modality or a whole trajectory point, then recover it. Pre-trained this way, TransferTraj reports average gains of 7.94% to 20.18% over the state-of-the-art baselines on trajectory prediction, trajectory recovery, and origin-destination travel-time estimation without task-specific retraining, and an 83.70% improvement in zero-shot region transfer on trajectory prediction against its configured baselines.

Load-bearing premise

The pre-trained text embeddings for POIs and road segments must be semantically consistent across cities and languages, so that the same kind of place maps to a similar vector regardless of region.

Editorial extensions

If this is right

  • A single pre-trained TransferTraj can serve trajectory prediction, trajectory recovery, and origin-destination travel-time estimation without retraining prediction heads, because all three tasks share the masking/recovery interface.
  • Zero-shot region transfer becomes practical: the paper reports an 83.70% gain over the SOTA baseline on trajectory prediction when the model moves between Chengdu, Xi'an, and Porto with no target-region training.
  • Few-shot region transfer with 5,000 target-region trajectories improves further, with reported gains of 33.68%, 18.08%, and 13.07% on the three tasks.
  • Deployment cost drops because the model is trained once and is lightweight: the efficiency study shows training time and memory at or below RNN-based baselines while avoiding per-task and per-region retraining.
  • The paper's stated limitation is that classification tasks with region-dependent output spaces, such as trajectory-user linking, are not covered by the proposed scheme.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the text-embedding invariance assumption holds, TransferTraj should also transfer across languages and map-data providers, since only the POI and road descriptions change; this is a testable extension the paper does not run.
  • Editorial inference: the masking/recovery interface is general enough that other generative spatiotemporal tasks, such as en-route arrival-time updates or destination prediction, could be expressed as additional masking patterns, though the paper does not demonstrate this.
  • Editorial inference: because region transfer is anchored in a pretrained text encoder, ablating that encoder (e.g., using a geographic-domain-specific or multilingual encoder) would separate the contribution of the architecture from the contribution of the embeddings.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper introduces TransferTraj, a vehicle GPS trajectory model designed to transfer across geographic regions and across generative tasks without retraining. The architecture combines a Region-Transferable Trajectory Encoder (RTTE), which fuses spatial, temporal, POI, and road-network modalities, with a task-transferable input-output scheme based on masking and recovering modalities or whole trajectory points. The authors report experiments on three datasets (Chengdu, Xi'an, Porto) for trajectory prediction, trajectory recovery, and origin-destination travel time estimation, under in-domain, task-transfer, zero-shot region-transfer, and few-shot region-transfer settings. The paper claims that pre-training alone outperforms state-of-the-art baselines by 7.94% to 20.18%, and that zero-shot region transfer improves trajectory prediction by 83.70% and OD TTE by 10.88%.

Significance. If the transfer claims survive a fair experimental protocol, TransferTraj is a plausible step toward a single trajectory model that can be reused across regions and tasks. The architecture is original in combining relative spatial information via a learnable rotation matrix, a spatial-context mixture-of-experts layer, and a unified masking scheme that covers multiple generative tasks. The paper includes extensive experiments, ablations isolating TRIE, SC-MoE, POI, and road-network components, and a hyperparameter analysis. The main value of the paper would be the demonstration that a single pre-trained model can handle several trajectory tasks and unseen regions without task-specific retraining; however, the evidence for the strongest forms of this claim is weakened by the experimental protocol issues and the narrow notion of task transfer used in the evaluation.

major comments (4)
  1. [Section 5.1, Table 4] The zero-shot trajectory-prediction comparison is not apples-to-apples. Every baseline in Table 4 is in the '(wo ft)' configuration, meaning its trajectory encoder is frozen and only a prediction head is fine-tuned, while the row labeled 'TransferTraj' carries no suffix. If that row is the fully fine-tuned model, then the headline 83.70% improvement over START (wo ft) largely reflects the fact that full fine-tuning is much stronger than frozen-encoder fine-tuning (compare START (wo ft) RMSE 1406.06 and START RMSE 319.00 on Xi'an in Table 1). If the row is instead the 'wo ft' variant, it should be labeled as such and the text should state this explicitly. The zero-shot claim should be recomputed against fully fine-tuned baselines, as is done for OD TTE in Table 5.
  2. [Section 4.2 and Appendix C] The evidence for task transferability is partly circular. The pre-training objective in Section 4.2 is defined as masking and recovering either a modality or a whole trajectory point, and Appendix C shows that the three target tasks are encoded as exactly these masking operations: TP masks future points, TR masks points at sampling gaps, and OD TTE masks the temporal modality of the last point. Hence the 'without retraining' results in Tables 1-3 measure performance on instances of the pre-training distribution, not on a task structure that was held out during pre-training. The paper should scope the task-transferability claim accordingly, and ideally test on at least one generative task whose input-output structure was not anticipated in the pre-training scheme.
  3. [Section 4.1.1] The region-transfer mechanism depends on an external text-embedding model for POI and road descriptions, but the paper does not state which OpenAI embedding model or version was used, whether it was frozen, or how language and geographic biases are controlled across Chinese and Portuguese text. The claim that the POI and road modalities provide 'region-independent semantic information' is load-bearing for zero-shot region transfer, yet no sensitivity analysis is provided. Please report the model/version, the language handling, and ideally an ablation using a different text encoder.
  4. [Section 5.1, Tables 1-3] The claimed average improvements of 20.18%, 17.87%, and 7.94% over START, MM-STGED, and DOT are not reproducible from the reported tables under standard aggregation. For example, taking the per-dataset RMSE improvement of TransferTraj (wo ft) over START in Table 1 and averaging the three datasets gives roughly 24.8%, not 20.18%; similar discrepancies appear for the other two tasks. Please specify the exact aggregation formula and, if MAE or MAPE or pooled RMSE is used, state this explicitly.
minor comments (5)
  1. [Appendix B.2 and Appendix G] Appendix B.2 states that the optimal hidden dimension is d=128, while Appendix G and Table 15 indicate that the optimal value of d is 256. Please reconcile this contradiction, since the reported results depend on the actual configuration.
  2. [Table 16] Table 16 appears to lack column headers and does not specify units for model size, training time, and testing time; the nine numeric entries per row are otherwise ambiguous.
  3. [Table 15] The SC-MoE hyperparameter rows are inconsistently labeled: 'c3' appears twice as 'c3 :k=4,C=6' and 'c3 :k=1,C=8', and the numbering c1,...,c6 does not match the c7 mentioned in Figure 5. Please correct the labels.
  4. [References] The t2vec method is cited as reference [9] in Table 1 and as reference [18] in the related-work discussion; both citations point to different entries. Please unify the reference labels.
  5. [Appendix D, Eq. (8)] The derivation of the relative rotation uses the notation '(xi,yi)-(xj,yj)' without defining componentwise subtraction; please clarify that this is the intended operation and that the trigonometric identities justify the block-diagonal result.
Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The model introduces no new physical or conceptual entities beyond standard neural network components. The free parameters are hyperparameters and hand-set context thresholds; the main external dependency is the OpenAI text-embedding model, whose exact version is not specified.

free parameters (6)
  • distance thresholds φpoi_dist and φroad_dist = 100 meters each
    Chosen by hand (Appendix B.2); determine which POIs and road segments enter each trajectory point context, directly shaping the spatial context modalities that drive region transfer.
  • hidden dimension d = reported as 128 in B.2, but 256 reported as optimal in Appendix G
    Tuned hyperparameter affecting model capacity; the conflicting reported values are an internal inconsistency.
  • number of stacked layers L = 2
    Optimal value from hyperparameter analysis (Appendix G), used in all experiments.
  • number of experts C = 8
    Optimal value from hyperparameter analysis (Appendix G), controls the SC-MoE capacity.
  • top-k routing k = 4
    Optimal value from hyperparameter analysis (Appendix G), controls expert sparsity.
  • pretraining epochs = 30
    Chosen in the experimental setting (Section 5); affects how well the masked-recovery objective is learned.
assumptions (4)
  • standard math Rotary position embeddings preserve relative positional information through rotation matrices, enabling the TRIE module to encode relative spatial coordinates.
    The TRIE design and Appendix D's proof rely on the angle-addition property of rotation matrices, a standard result from RoFormer.
  • domain assumption The pre-trained text embedding model yields transferable semantic representations for POIs and roads across geographic regions.
    Section 4.1.1 uses an external text embedding model as the source of region-independent semantic context; the paper provides no analysis of cross-lingual or cross-region embedding alignment.
  • domain assumption Vehicle movement patterns are governed by local spatial context, so a mixture-of-experts can share experts across regions when context is similar.
    SC-MoE is motivated by this observation (Section 4.1.3); the expert activation analysis in Figure 2 provides only indirect evidence within one city.
  • domain assumption Representing points relative to the first point of a trajectory preserves spatial scale information across regions better than absolute normalization.
    Section 4.1.1 asserts this design choice without a comparative analysis against other coordinate standardization methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability." pith.science (2026). https://pith.science/paper/4WQCQHXN

@misc{pith2026250512672,
  author       = {Pith},
  title        = {Pith review of: TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4WQCQHXN}},
  note         = {Machine review of arXiv:2505.12672}
}
read the original abstract

Vehicle GPS trajectories provide valuable movement information that supports various downstream tasks and applications. A desirable trajectory learning model should be able to transfer across regions and tasks without retraining, avoiding the need to maintain multiple specialized models and subpar performance with limited training data. However, each region has its unique spatial features and contexts, which are reflected in vehicle movement patterns and difficult to generalize. Additionally, transferring across different tasks faces technical challenges due to the varying input-output structures required for each task. Existing efforts towards transferability primarily involve learning embedding vectors for trajectories, which perform poorly in region transfer and require retraining of prediction modules for task transfer. To address these challenges, we propose TransferTraj, a vehicle GPS trajectory learning model that excels in both region and task transferability. For region transferability, we introduce RTTE as the main learnable module within TransferTraj. It integrates spatial, temporal, POI, and road network modalities of trajectories to effectively manage variations in spatial context distribution across regions. It also introduces a TRIE module for incorporating relative information of spatial features and a spatial context MoE module for handling movement patterns in diverse contexts. For task transferability, we propose a task-transferable input-output scheme that unifies the input-output structure of different tasks into the masking and recovery of modalities and trajectory points. This approach allows TransferTraj to be pre-trained once and transferred to different tasks without retraining. Extensive experiments on three real-world vehicle trajectory datasets under task transfer, zero-shot, and few-shot region transfer, validating TransferTraj's effectiveness.

Figures

Figures reproduced from arXiv: 2505.12672 by the authors.

Figure 1
Figure 1. The framework of TransferTraj. trajectory points, preventing the model from becoming biased towards specific regions. The SC-MoE block dynamically selects appropriate movement pattern experts based on the local spatial context, allowing the model to adapt flexibly to diverse contexts. The task-transferable input-output scheme equips TransferTraj with the flexibility to handle various tasks. It standardizes the input… view at source ↗
Figure 2
Figure 2. Expect activation distrubition. Analysis of the SC-MoE. We analyze the experts activated across different spatial contexts by grouping trajectory points based on the density of nearby POIs and road segments. Specifically, we classify the local spatial context into three categories: high-density regions (> 15 POIs and road segments), medium-density regions (5 ∼ 15), and low-density regions (< 5). For each category, w… view at source ↗
Figure 3
Figure 3. Hyperparameter analysis of the d. 1 2 3 4 5 6 150 200 250 300 RMSE RMSE MAE 150 200 250 MAE Chengdu 1 2 3 4 5 6 200 250 300 RMSE RMSE MAE 150 200 250 MAE Xi'an 1 2 3 4 5 6 200 250 RMSE RMSE MAE 140 160 180 200 MAE Porto [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Hyperparameter analysis of the L. G Hyperparameter Study We analyze three key hyperparameters: the hidden state dimension d of the model, the number of block L of stacked TRIE and SC-MoE, and the value of k and the number of experts C in the SC-MoE. The values of these…
Figure 5
Figure 5. Figure 5: Hyperparameter analysis of the value of k and the number of expert C. harder to train. The optimal value for d is 256; smaller values limit the model’s learning capability, while larger values offer minimal performance gains and decrease computational efficiency. For t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 31 canonical work pages

  1. [1]

    Contrastive trajectory similarity learning with dual-feature attention

    Yanchuan Chang, Jianzhong Qi, Yuxuan Liang, and Egemen Tanin. Contrastive trajectory similarity learning with dual-feature attention. InICDE, pages 2933–2945, 2023

    work page 2023

  2. [2]

    Xgboost: A scalable tree boosting system

    Tianqi Chen and Carlos Guestrin. Xgboost: A scalable tree boosting system. InProceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pages 785–794, 2016

    2016

  3. [3]

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey E. Hinton. A simple framework for contrastive learning of visual representations. InICML, volume 119, pages 1597–1607, 2020

    work page 2020

  4. [4]

    Robust road network representation learning: When traffic patterns meet traveling semantics

    Yile Chen, Xiucheng Li, Gao Cong, Zhifeng Bao, Cheng Long, Yiding Liu, Arun Kumar Chandran, and Richard Ellison. Robust road network representation learning: When traffic patterns meet traveling semantics. InCIKM, pages 211–220, 2021

    work page 2021

  5. [5]

    Rntrajrec: Road network enhanced trajectory recovery with spatial-temporal transformer

    Yuqi Chen, Hanyuan Zhang, Weiwei Sun, and Baihua Zheng. Rntrajrec: Road network enhanced trajectory recovery with spatial-temporal transformer. In2023 IEEE 39th International Conference on Data Engineering (ICDE), pages 829–842. IEEE, 2023

    work page 2023

  6. [6]

    Discovering popular routes from trajectories

    Zaiben Chen, Heng Tao Shen, and Xiaofang Zhou. Discovering popular routes from trajectories. In2011 IEEE 27th International Conference on Data Engineering, pages 900–911. IEEE, 2011

    work page 2011

  7. [7]

    Empirical evaluation of gated recurrent neural networks on sequence modeling.arXiv preprint arXiv:1412.3555, 2014

    Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, and Yoshua Bengio. Empirical evaluation of gated recurrent neural networks on sequence modeling.arXiv preprint arXiv:1412.3555, 2014

    arXiv 2014

  8. [8]

    Network-less trajectory imputation

    Mohamed M Elshrif, Keivin Isufaj, and Mohamed F Mokbel. Network-less trajectory imputation. In Proceedings of the 30th International Conference on Advances in Geographic Information Systems, pages 1–10, 2022

    work page 2022

Show all 41 references
  1. [9]

    Ziquan Fang, Yuntao Du, Xinjun Zhu, Danlei Hu, Lu Chen, Yunjun Gao, and Christian S. Jensen. Spatio- temporal trajectory similarity learning in road networks. InKDD, pages 347–356, 2022

    2022

  2. [10]

    Deepmove: Predicting human mobility with attentional recurrent networks

    Jie Feng, Yong Li, Chao Zhang, Funing Sun, Fanchao Meng, Ang Guo, and Depeng Jin. Deepmove: Predicting human mobility with attentional recurrent networks. InWWW, pages 1459–1468, 2018

    2018

  3. [11]

    TremBR: Exploring road networks for trajectory representation learning.ACM Trans

    Tao-Yang Fu and Wang-Chien Lee. TremBR: Exploring road networks for trajectory representation learning.ACM Trans. Intell. Syst. Technol., 11(1):10:1–10:25, 2020

    2020

  4. [12]

    Reducing the dimensionality of data with neural networks

    Geoffrey E Hinton and Ruslan R Salakhutdinov. Reducing the dimensionality of data with neural networks. science, 313(5786):504–507, 2006

    2006

  5. [13]

    Long short-term memory.Neural computation, 9(8):1735–1780, 1997

    Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory.Neural computation, 9(8):1735–1780, 1997

    1997

  6. [14]

    A learning-based method for computing shortest path distances on road networks

    Shuai Huang, Yong Wang, Tianyu Zhao, and Guoliang Li. A learning-based method for computing shortest path distances on road networks. In2021 IEEE 37th International Conference on Data Engineering (ICDE), pages 360–371. IEEE, 2021

    2021

  7. [15]

    Self-supervised trajectory representation learning with temporal regularities and travel semantics

    Jiawei Jiang, Dayan Pan, Houxing Ren, Xiaohan Jiang, Chao Li, and Jingyuan Wang. Self-supervised trajectory representation learning with temporal regularities and travel semantics. InICDE, pages 843–855, 2023

    2023

  8. [16]

    A unified neural network approach for estimating travel time and distance for a taxi trip.arXiv preprint arXiv:1710.04350, 2017

    Ishan Jindal, Xuewen Chen, Matthew Nokleby, Jieping Ye, et al. A unified neural network approach for estimating travel time and distance for a taxi trip.arXiv preprint arXiv:1710.04350, 2017

    2017 arXiv

  9. [17]

    HST-LSTM: A hierarchical spatial-temporal long-short term memory network for location prediction

    Dejiang Kong and Fei Wu. HST-LSTM: A hierarchical spatial-temporal long-short term memory network for location prediction. InIJCAI, pages 2341–2347, 2018. 10

    2018

  10. [18]

    Jensen, and Wei Wei

    Xiucheng Li, Kaiqi Zhao, Gao Cong, Christian S. Jensen, and Wei Wei. Deep representation learning for trajectory similarity computation. InICDE, pages 617–628

  11. [19]

    Multi-task representation learning for travel time estimation

    Yaguang Li, Kun Fu, Zheng Wang, Cyrus Shahabi, Jieping Ye, and Yan Liu. Multi-task representation learning for travel time estimation. InKDD, pages 1695–1704, 2018

    2018

  12. [20]

    Learnable fourier features for multi-dimensional spatial positional encoding.Advances in Neural Information Processing Systems, 34:15816–15829, 2021

    Yang Li, Si Si, Gang Li, Cho-Jui Hsieh, and Samy Bengio. Learnable fourier features for multi-dimensional spatial positional encoding.Advances in Neural Information Processing Systems, 34:15816–15829, 2021

    2021

  13. [21]

    Modeling trajectories with neural ordinary differential equations

    Yuxuan Liang, Kun Ouyang, Hanshu Yan, Yiwei Wang, Zekun Tong, and Roger Zimmermann. Modeling trajectories with neural ordinary differential equations. InIJCAI, pages 1498–1504, 2021

    2021

  14. [22]

    Jensen, and Youfang Lin

    Yan Lin, Huaiyu Wan, Shengnan Guo, Jilin Hu, Christian S. Jensen, and Youfang Lin. Pre-training general trajectory embeddings with maximum multi-view entropy coding.IEEE Trans. Knowl. Data Eng., pages 1–15, 2023

    2023

  15. [23]

    Pre-training context and time aware location embeddings from spatial-temporal trajectories for user next location prediction

    Yan Lin, Huaiyu Wan, Shengnan Guo, and Youfang Lin. Pre-training context and time aware location embeddings from spatial-temporal trajectories for user next location prediction. InAAAI, pages 4241–4248, 2021

    2021

  16. [24]

    Yan Lin, Huaiyu Wan, Jilin Hu, Shengnan Guo, Bin Yang, Youfang Lin, and Christian S. Jensen. Origin- destination travel time oracle for map-based services.PACMMOD, 1(3):217:1–217:27, 2023

    2023

  17. [25]

    Predicting human mobility via attentive convolutional network

    Congcong Miao, Ziyan Luo, Fengzhu Zeng, and Jilong Wang. Predicting human mobility via attentive convolutional network. InWSDM, pages 438–446, 2020

    2020

  18. [26]

    Mtrajrec: Map-constrained trajectory recovery via seq2seq multi-task learning

    Huimin Ren, Sijie Ruan, Yanhua Li, Jie Bao, Chuishi Meng, Ruiyuan Li, and Yu Zheng. Mtrajrec: Map-constrained trajectory recovery via seq2seq multi-task learning. InProceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 1410–1419, 2021

    2021

  19. [27]

    Outrageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprint arXiv:1701.06538, 2017

    Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprint arXiv:1701.06538, 2017

    2017 arXiv

  20. [28]

    Roformer: Enhanced transformer with rotary position embedding.Neurocomputing, 568:127063, 2024

    Jianlin Su, Murtadha Ahmed, Yu Lu, Shengfeng Pan, Wen Bo, and Yunfeng Liu. Roformer: Enhanced transformer with rotary position embedding.Neurocomputing, 568:127063, 2024

    2024

  21. [29]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. InNeurIPS, pages 5998–6008, 2017

    2017

  22. [30]

    Pre-training time-aware location embeddings from spatial-temporal trajectories.IEEE Trans

    Huaiyu Wan, Yan Lin, Shengnan Guo, and Youfang Lin. Pre-training time-aware location embeddings from spatial-temporal trajectories.IEEE Trans. Knowl. Data Eng., 34(11):5510–5523, 2022

    2022

  23. [31]

    A simple baseline for travel time estimation using large-scale trip data

    Hongjian Wang, Yu-Hsuan Kuo, Daniel Kifer, and Zhenhui Li. A simple baseline for travel time estimation using large-scale trip data. InSIGSPATIAL, pages 61:1–61:4, 2016

    2016

  24. [32]

    Deep trajectory recovery with fine-grained calibration using kalman filter.IEEE Transactions on Knowledge and Data Engineering, 33(3):921–934, 2019

    Jingyuan Wang, Ning Wu, Xinxi Lu, Wayne Xin Zhao, and Kai Feng. Deep trajectory recovery with fine-grained calibration using kalman filter.IEEE Transactions on Knowledge and Data Engineering, 33(3):921–934, 2019

    2019

  25. [33]

    Ptr: A pre-trained language model for trajectory recovery.arXiv preprint arXiv:2410.14281, 2024

    Tonglong Wei, Yan Lin, Youfang Lin, Shengnan Guo, Jilin Hu, Gao Cong, and Huaiyu Wan. Ptr: A pre-trained language model for trajectory recovery.arXiv preprint arXiv:2410.14281, 2024

    2024 arXiv

  26. [34]

    Micro-macro spatial- temporal graph-based encoder-decoder for map-constrained trajectory recovery.IEEE Transactions on Knowledge and Data Engineering, 2024

    Tonglong Wei, Youfang Lin, Yan Lin, Shengnan Guo, Lan Zhang, and Huaiyu Wan. Micro-macro spatial- temporal graph-based encoder-decoder for map-constrained trajectory recovery.IEEE Transactions on Knowledge and Data Engineering, 2024

    2024

  27. [35]

    Deepeta: A spatial-temporal sequential neural network model for estimating time of arrival in package delivery system

    Fan Wu and Lixia Wu. Deepeta: A spatial-temporal sequential neural network model for estimating time of arrival in package delivery system. InAAAI, pages 774–781, 2019

    2019

  28. [36]

    Modeling trajectories with recurrent neural networks

    Hao Wu, Ziyang Chen, Weiwei Sun, Baihua Zheng, and Wei Wang. Modeling trajectories with recurrent neural networks. InIJCAI, pages 3083–3090, 2017

    2017

  29. [37]

    PreCLN: Pretrained-based contrastive learning network for vehicle trajectory prediction.WWW, 26(4):1853–1875, 2023

    Bingqi Yan, Geng Zhao, Lexue Song, Yanwei Yu, and Junyu Dong. PreCLN: Pretrained-based contrastive learning network for vehicle trajectory prediction.WWW, 26(4):1853–1875, 2023

    2023

  30. [38]

    Sean Bin Yang, Jilin Hu, Chenjuan Guo, Bin Yang, and Christian S. Jensen. Lightpath: Lightweight and scalable path representation learning. InKDD, pages 2999–3010, 2023. 11

    2023

  31. [39]

    Trajectory clustering via deep representation learning

    Di Yao, Chao Zhang, Zhihua Zhu, Jian-Hui Huang, and Jingping Bi. Trajectory clustering via deep representation learning. InIJCNN, pages 3880–3887, 2017

    2017

  32. [40]

    A survey of traffic prediction: from spatio-temporal data to intelligent transportation.Data Sci

    Haitao Yuan and Guoliang Li. A survey of traffic prediction: from spatio-temporal data to intelligent transportation.Data Sci. Eng., 6(1):63–85, 2021

    2021

  33. [41]

    Effective travel time estimation: When historical trajectories over road networks matter

    Haitao Yuan, Guoliang Li, Zhifeng Bao, and Ling Feng. Effective travel time estimation: When historical trajectories over road networks matter. InSIGMOD, pages 2135–2149, 2020. 12 Table 7: Statistics of datasets. Dataset Chengdu Xian Porto # Trajectories 140,000 210,000 323, 4...

    2020

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.