Pith. sign in

REVIEW 4 major objections 4 minor 2 cited by

RED: Effective Trajectory Representation Learning with Comprehensive Information

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper claims that RED, a self-supervised trajectory encoder with road-aware masking and joint spatial-temporal-user embeddings, beats nine existing methods on four downstream tasks.

desk verdict RED is a solid self-supervised trajectory representation learning framework with real gains on the three headline tasks, but its universal superiority claim is contradicted by its own Table 6. read the letter →

arxiv 2411.15096 v2 pith:3XPJRERI submitted 2024-11-22 cs.LG cs.AI

Silin Zhou , Shuo Shang , Lisi Chen , Christian S. Jensen , Panos Kalnis This is my paper · ORCID
classification cs.LGcs.AI
keywords trajectoryrepresentationlearningself-supervisedmaskedautoencoderroad-awaremaskingspatial-temporalembeddingtraveltimeestimationsimilarityclassification
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

The paper sets out to show that trajectory representation learning can be made markedly more accurate by using the information a trajectory already carries, including road, user, spatial, temporal, travel, and movement signals, rather than by relying on stronger data augmentation. It introduces RED, a Transformer-based masked autoencoder that masks only the less important road segments of a path trajectory, embeds spatial, temporal, and user information jointly, and trains with two objectives: predicting the next road segment and reconstructing the full trajectory. On three real-world datasets and against nine trajectory representation baselines, RED reports average accuracy improvements over the best baseline of 7.03% for travel time estimation, 12.11% for trajectory classification, and 20.02% for trajectory similarity computation. The intended message is that comprehensive information, not augmentation, is what current trajectory representation learning is missing.

What carries the argument

The load-bearing mechanism is road-aware masking: each road segment is classified as key or mask by comparing its sampling-point count and its length with the dataset-wide averages, and the encoder of the masked autoencoder only sees the key path while the decoder reconstructs the entire trajectory. This is supported by the spatial-temporal-user joint embedding, which sums a graph attention network (GAT)-based spatial encoding, a learned time-and-segment-type encoding, and a user embedding, and by virtual tokens [START], [END], and [EXTRACT] that fix input-output misalignment; the [EXTRACT] token's output is the trajectory vector. A time-distance enhanced attention adds time-interval and distance-interval correlations to the self-attention scores.

What would settle it

On a dataset where all segments have comparable length and sampling counts, road-aware masking becomes indistinguishable from random masking; if RED still outperforms baselines by a large margin there, the gains come from another component, and if the margin vanishes, the masking heuristic is the source.

Watch

Extended reading notes

Core claim

The central discovery is that a road-aware masking strategy, which never masks road segments that are 'hot' (more sampling points than the average segment) or 'long' (longer than the average segment), preserves the semantic backbone of a trajectory and yields vector representations that transfer better to downstream tasks than representations trained with random masking. The paper bundles this with a spatial-temporal-user joint embedding, a dual objective of next-segment prediction and trajectory reconstruction, and a time-distance enhanced attention mechanism, and the empirical claim is that every component contributes. The largest reported gains appear on trajectory similarity computation, where the learned vectors replace quadratic dynamic-programming comparisons with linear vector operations.

Load-bearing premise

The claim rests on the premise that road segments with above-average sampling points or above-average length are the semantically important ones, so a mask that preserves them keeps the trajectory's meaning intact without leaking test-set information.

Editorial extensions

If this is right

  • With RED's pre-trained vectors, computing the similarity of two trajectories costs O(l) instead of the quadratic cost of dynamic-programming measures, so large-scale retrieval and clustering become practical at higher accuracy.
  • Because road-aware masking adapts the mask per trajectory without tuning a mask ratio, RED can be applied to new road-network datasets without the per-dataset mask-ratio search that random masking requires.
  • The encoder alone is used at inference, with complete trajectories as input, so fine-tuning for travel time estimation or classification does not change the representation architecture.
  • The method is designed for road-network trajectories and would need adjustment for POI, animal, or pedestrian trajectories, and it targets trajectory-level tasks rather than road-level tasks such as flow estimation.

Reading between the lines

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

  • If the hot/long heuristic is the real source of the gains, its advantage should shrink on datasets where segment lengths and sampling counts are nearly uniform; this is directly testable by comparing RED against a version with oracle key paths chosen from ground-truth travel time.
  • The next-segment prediction objective makes trajectories resemble a path-level language model, which suggests that larger pre-training corpora and scaling laws, rather than better augmentations, may drive further gains in trajectory representation learning.
  • The paper does not state whether the hot/long aggregate statistics are computed on the training split only; if they are computed on the full dataset, the mask itself could carry test-set information and inflate the reported improvements.
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 / 4 minor

Summary. The paper proposes RED, a self-supervised Transformer-based trajectory representation learning framework. RED's main components are a road-aware masking strategy that preserves segments deemed hot or long, a spatial-temporal-user joint embedding, dual-objective learning (next-segment prediction and full-trajectory reconstruction), and a time-distance enhanced attention module. The authors evaluate RED on Porto, Rome, and Chengdu across four downstream tasks (travel time estimation, trajectory classification, trajectory similarity computation, and most-similar trajectory retrieval), compare with nine TRL baselines and seven heuristic similarity measures, and report ablations and an efficiency study. The central claim is that RED outperforms all existing methods in accuracy across tasks and datasets, with average improvements of 7.03%, 12.11%, and 20.02% over the best baseline on three of the tasks.

Significance. If the reported results are reproducible, RED is a meaningful contribution to trajectory representation learning: it combines road, user, spatial, temporal, and movement information in a single framework, uses a masked-autoencoder design adapted to trajectories, and provides an efficiency analysis plus a generalization experiment on Geolife. The paper also makes its code and data available. However, the headline claim of universal superiority is not supported by the paper's own retrieval results, and two recent baselines discussed in the related work are absent from the experiments. The contribution is nevertheless valuable as an architecture and empirical study, provided the claims are scoped and statistical evidence is added.

major comments (4)
  1. [Section 1 and Section 5.2, Table 6] The abstract and introduction state that "RED outperforms all existing methods in terms of accuracy across the tasks and datasets," but Table 6 contradicts this: on Porto most-similar trajectory retrieval, START is better than RED in every configuration (e.g., MR 1.232 vs. 1.420 at p=0.1 on 10k; 8.831 vs. 10.77 at p=0.4 on 100k). Section 5.2 itself acknowledges that RED is "slightly lower than START on Porto." The universal superiority claim is therefore false under the paper's own evaluation unless the claim is explicitly restricted to the three tasks named in the abstract's improvement numbers. Please revise the claim accordingly and discuss the Porto retrieval result honestly in the main text.
  2. [Section 5.1 baselines vs. Section 6 related work] JCLRNT and LightPath are described in Section 6 as recent state-of-the-art Transformer-based TRL methods, yet neither appears in the experimental comparison of Section 5.1. The abstract claims improvement over "all existing methods," which is unsupported when two recent methods are omitted. Please include these baselines in the experiments or explicitly scope the claim to the nine compared methods; the current wording overstates the evidence.
  3. [Section 4.1 road-aware masking] The hot/long thresholds are defined by comparing each segment's sampling-point count and length with "the average over the road segments," but the paper does not state whether these averages are computed on the training split only. If the full dataset, including test trajectories, is used to set the thresholds, the masking procedure leaks test information into pretraining. Please clarify the split used and, ideally, recompute thresholds on the training split; additionally, report sensitivity of the downstream metrics to the threshold choice, since this heuristic is load-bearing for the claimed benefit of road-aware masking.
  4. [Tables 3–6 and 10] No error bars, confidence intervals, or repeated runs are reported for any downstream result. Several claimed improvements are small (e.g., Chengdu classification accuracy improves by 1.21% and F1 by 0.68% in Table 4), so without variance estimates it is impossible to judge whether these differences are significant. Please run multiple seeds and report means with standard deviations or significance tests, particularly for the smaller improvements.
minor comments (4)
  1. [Section 5.2] The text contains a typo: "Tabel 2" should be "Table 2."
  2. [Section 4.2, Eq. (4)] The dimension d of the time encoding t_i is not defined; it should be stated explicitly (presumably d = l, the embedding dimension), and the dimensions of FC1 and FC2 should be specified so that the concatenation in Eq. (4) is unambiguous.
  3. [Section 4.4.2] The phrase "two distant segments have has lower correlation" contains a grammatical error; it should read "two distant segments have lower correlation."
  4. [Section 5.3, Table 9] The Porto and Rome rows in Table 9 are formatted as one continuous line in the text; please separate them clearly for readability.

Circularity Check

1 steps flagged · score 6.0 of 10

Travel-time gains reduce to a segment-average-travel-time input feature; classification/similarity results are independent.

  1. fitted input called prediction [Section 4.2 (Spatial Encoding) and Section 5.1 (Downstream Task Settings)]
    "We feed multiple attributes of a segment as the initial input feature of GAT, including the maximum speed limit, average travel time, segment direction, out-degree, in-degree, segment length, and segment type. ... For travel time estimation, we remove all time information, including the time encoding as well as the time correlation, except for the departure time, to avoid time information leakage."

    The downstream target is the total travel time of a trajectory, i.e., the time to traverse a sequence of road segments. The spatial encoder is given the 'average travel time' of each road segment as an input attribute, so a readout or weighted sum of these per-segment averages is already a strong, almost construction-level predictor of the target. The downstream setup explicitly removes temporal encodings and time correlation to prevent leakage, but it does not state that the average-travel-time segment attribute is also removed. The reported travel-time improvement is therefore partly a fitted-input prediction rather than an independent prediction from trajectory structure.

full rationale

The self-supervised pretraining objectives (masked trajectory reconstruction and next-segment prediction) are not definitionally tied to the downstream classification and similarity metrics; those evaluations are held out and provide independent evidence for RED's representation quality. No load-bearing self-citation chain is present: citations to prior work by overlapping authors appear in related work and are not used to justify the method's validity. The one significant circular step is in the travel-time estimation setup: the GAT spatial encoding uses 'average travel time' per road segment as an input feature, and the downstream task predicts the travel time of a trajectory made of those segments. Since the paper's leakage-avoidance procedure removes only temporal encodings, not this per-segment target statistic, the travel-time results are substantially forced by the input. A separate correctness issue, not counted as circularity, is that Table 6 shows RED losing to START on Porto for most-similar retrieval, contradicting the universal 'outperforms all' claim; this is an overclaim rather than a circular derivation. Overall, the classification and similarity claims remain independent, so the paper is only partially circular.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The central claims rest on standard deep learning machinery, the domain assumption that road-network path trajectories are a faithful and sufficient representation, and the ad hoc hot/long heuristic for choosing key paths. The free parameters are standard hyperparameters plus the data-derived threshold; none of these are fitted to the downstream labels, so the burden is moderate.

free parameters (5)
  • lambda1 = 0.1
    Weight balancing next-segment prediction and trajectory reconstruction in Eq. (2); chosen by hand.
  • lambda2 = 0.5
    Mix ratio for time vs distance in the enhanced attention (Eq. 7); tuned via Figure 6.
  • embedding dimension = 128
    Selected from {16,32,64,128,256} based on validation HR@10; see Figure 7.
  • number of encoder/decoder layers = 6
    Selected from {1,...,7} based on validation; see Figure 7.
  • hot/long threshold = dataset-wide average sampling count and length
    Threshold defining key paths in road-aware masking; computed from data statistics in Section 4.1.
assumptions (5)
  • standard math Transformer and GAT provide sufficiently expressive sequence and graph encoders for trajectory modeling
    Assumed as the architectural backbone; no formal guarantee is provided.
  • domain assumption Map matching faithfully converts GPS trajectories to road-segment path trajectories
    Section 2.1 uses FMM map matching [39] and treats the resulting path trajectories as the input representation.
  • ad hoc to paper Road segments with above-average sampling counts or lengths are the semantically important key paths
    Section 4.1 motivates this heuristic with Porto statistics and applies it to all datasets.
  • ad hoc to paper The transformation f(m) = 1/log(e + g(m)) captures the desired decay of attention with time/distance
    Section 4.4.2 introduces this form without comparison to alternatives.
  • ad hoc to paper A virtual [START] node connected to all road segments preserves road topology when predicting the first segment
    Section 4.4.1 augments the road graph with this node.
invented entities (1)
  • Virtual tokens [START], [END], [EXTRACT] and the virtual [START] graph node
    purpose: Fix input-output segment misalignment in next-segment prediction and provide a trajectory-level representation vector
    These are internal model components with no falsifiable prediction outside the paper's own experiments; their effectiveness is shown only through ablations in Table 10.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RED: Effective Trajectory Representation Learning with Comprehensive Information." pith.science (2026). https://pith.science/paper/3XPJRERI

@misc{pith2026241115096,
  author       = {Pith},
  title        = {Pith review of: RED: Effective Trajectory Representation Learning with Comprehensive Information},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3XPJRERI}},
  note         = {Machine review of arXiv:2411.15096}
}
read the original abstract

Trajectory representation learning (TRL) maps trajectories to vectors that can then be used for various downstream tasks, including trajectory similarity computation, trajectory classification, and travel-time estimation. However, existing TRL methods often produce vectors that, when used in downstream tasks, yield insufficiently accurate results. A key reason is that they fail to utilize the comprehensive information encompassed by trajectories. We propose a self-supervised TRL framework, called RED, which effectively exploits multiple types of trajectory information. Overall, RED adopts the Transformer as the backbone model and masks the constituting paths in trajectories to train a masked autoencoder (MAE). In particular, RED considers the moving patterns of trajectories by employing a Road-aware masking strategy} that retains key paths of trajectories during masking, thereby preserving crucial information of the trajectories. RED also adopts a spatial-temporal-user joint Embedding scheme to encode comprehensive information when preparing the trajectories as model inputs. To conduct training, RED adopts Dual-objective task learning}: the Transformer encoder predicts the next segment in a trajectory, and the Transformer decoder reconstructs the entire trajectory. RED also considers the spatial-temporal correlations of trajectories by modifying the attention mechanism of the Transformer. We compare RED with 9 state-of-the-art TRL methods for 4 downstream tasks on 3 real-world datasets, finding that RED can usually improve the accuracy of the best-performing baseline by over 5%.

Figures

Figures reproduced from arXiv: 2411.15096 by the authors.

Figure 1
Figure 1. Overall architecture of the RED framework. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Trajectory sample statistics of the Porto dataset. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. The enhanced Transformer of the encoder and decoder. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: Trajectory similarity computation time (in s) when [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Influence of the time-distance attention ratio (i.e., [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Effect of embedding dimension and model layers. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?

    cs.MM 2025-08 reject novelty 6.0 of 10

    Traj-MLLM claims a training-free MLLM framework outperforms task-specific trajectory models on four tasks, but the input may leak the target for travel time and destination prediction.

  2. TrajSceneLLM: A Multimodal Perspective on Semantic GPS Trajectory Analysis

    cs.CY 2025-06 conditional novelty 6.0 of 10

    TrajSceneLLM combines map images and LLM-generated text into embeddings that reach 86.8% accuracy on GeoLife travel mode identification, 2.4 points above the prior MASO-MSF method.

Reference graph

Works this paper leans on

54 extracted references · 46 canonical work pages · cited by 2 Pith papers

  1. [1]

    Helmut Alt. 2009. The Computational Geometry of Comparing Shapes.Efficient Algorithms, Essays Dedicated to Kurt Mehlhorn on the Occasion of His 60th Birthday 5760 (2009), 235–248

    work page 2009

  2. [2]

    Helmut Alt and Michael Godau. 1995. Computing the Fréchet distance between two polygonal curves. International Journal of Computational Geometry & Applications 5 (1995), 75–91

    work page 1995

  3. [3]

    Berndt and James Clifford

    Donald J. Berndt and James Clifford. 1994. Using Dynamic Time Warping to Find Patterns in Time Series. InAAAI. 359–370

    work page 1994

  4. [4]

    Besse, Brendan Guillouet, Jean-Michel Loubes, and François Royer

    Philippe C. Besse, Brendan Guillouet, Jean-Michel Loubes, and François Royer

    work page

  5. [5]

    Leo Breiman. 2001. Random Forests.Machine learning45, 1 (2001), 5–32

    work page 2001

  6. [6]

    Yanchuan Chang, Jianzhong Qi, Yuxuan Liang, and Egemen Tanin. 2023. Con- trastive Trajectory Similarity Learning with Dual-Feature Attention. InICDE. 2933–2945

    work page 2023

  7. [7]

    Zhengping Che, Sanjay Purushotham, Kyunghyun Cho, David Sontag, and Yan Liu. 2018. Recurrent neural networks for multivariate time series with missing values. Scientific reports8, 1 (2018), 6085

    work page 2018

  8. [8]

    Lei Chen and Raymond T. Ng. 2004. On The Marriage of Lp-norms and Edit Distance. InVLDB. 792–803

    work page 2004

Show all 54 references
  1. [9]

    Tamer Özsu, and Vincent Oria

    Lei Chen, M. Tamer Özsu, and Vincent Oria. 2005. Robust and Fast Similarity Search for Moving Object Trajectories. InSIGMOD. 491–502

    2005

  2. [10]

    Yile Chen, Xiucheng Li, Gao Cong, Zhifeng Bao, Cheng Long, Yiding Liu, Arun Kumar Chandran, and Richard Ellison. 2021. Robust Road Network Representation Learning: When Traffic Patterns Meet Traveling Semantics. In CIKM. 211–220

    2021

  3. [11]

    BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

    JacobDevlin,Ming-WeiChang,KentonLee,andKristinaToutanova.2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In NAACL. 4171–4186

    2019

  4. [12]

    Ziquan Fang, Yuntao Du, Lu Chen, Yujia Hu, Yunjun Gao, and Gang Chen. 2021. E2DTC: An End to End Deep Trajectory Clustering Framework via Self-Training. In ICDE. 696–707

    2021

  5. [13]

    Tao-Yang Fu and Wang-Chien Lee. 2020. Trembr: Exploring road networks for trajectory representation learning.ACM Transactions on Intelligent Systems and Technology11, 1 (2020), 1–25

    2020

  6. [14]

    Qiang Gao, Wei Wang, Kunpeng Zhang, Xin Yang, Congcong Miao, and Tianrui Li. 2022. Self-supervised representation learning for trip recommendation. Knowledge-Based Systems247 (2022), 108791

    2022

  7. [15]

    Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable Feature Learning for Networks. InKDD. 855–864

    2016

  8. [16]

    Peng Han, Jin Wang, Di Yao, Shuo Shang, and Xiangliang Zhang. 2021. A Graph-basedApproachforTrajectorySimilarityComputationinSpatialNetworks. In KDD. 556–564

    2021

  9. [17]

    Girshick

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross B. Girshick. 2022. Masked Autoencoders Are Scalable Vision Learners. InCVPR. 15979–15988

    2022

  10. [18]

    Hearst, S.T

    M.A. Hearst, S.T. Dumais, E. Osuna, J. Platt, and B. Scholkopf. 1998. Support vector machines.IEEE Intelligent Systems and their Applications13, 4 (1998), 18–28

    1998

  11. [19]

    Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. Neural Computation9, 8 (1997), 1735–1780

    1997

  12. [20]

    Jiawei Jiang, Dayan Pan, Houxing Ren, Xiaohan Jiang, Chao Li, and Jingyuan Wang. 2023. Self-supervised Trajectory Representation Learning with Temporal Regularities and Travel Semantics. InICDE. 843–855

    2023

  13. [21]

    Brubaker

    Seyed Mehran Kazemi, Rishab Goel, Sepehr Eghbali, Janahan Ramanan, Jaspreet Sahota, Sanjay Thakur, Stella Wu, Cathal Smyth, Pascal Poupart, and Marcus A. Brubaker. 2019. Time2vec: Learning a vector representation of time. InarXiv preprint. https://arxiv.org/pdf/1907.05321

    2019 arXiv

  14. [22]

    Ai-Te Kuo, Haiquan Chen, and Wei-Shinn Ku. 2023. BERT-Trip: Effective and Scalable Trip Representation using Attentive Contrast Learning. InICDE. 612–623

    2023

  15. [23]

    Jensen, and Wei Wei

    Xiucheng Li, Kaiqi Zhao, Gao Cong, Christian S. Jensen, and Wei Wei. 2018. Deep Representation Learning for Trajectory Similarity Computation. InICDE. 617–628

    2018

  16. [24]

    Yuxuan Liang, Kun Ouyang, Yiwei Wang, Xu Liu, Hongyang Chen, Junbo Zhang, Yu Zheng, and Roger Zimmermann. 2022. TrajFormer: Efficient Trajectory Classification with Transformers. InCIKM. 1229–1237

    2022

  17. [25]

    Yuxuan Liang, Kun Ouyang, Hanshu Yan, Yiwei Wang, Zekun Tong, and Roger Zimmermann. 2021. Modeling Trajectories with Neural Ordinary Differential Equations. InIJCAI. 1498–1504

    2021

  18. [26]

    Uni- fied route representation learning for multi-modal transportation recommendation with spatiotemporal pre-training.The VLDB Journal32, 2 (2023), 325–342

    HaoLiu,JindongHan,YanjieFu,YanyanLi,KaiChen,andHuiXiong.2023. Uni- fied route representation learning for multi-modal transportation recommendation with spatiotemporal pre-training.The VLDB Journal32, 2 (2023), 325–342

    2023

  19. [27]

    Hao Liu, Wenzhao Jiang, Shui Liu, and Xi Chen. 2023. Uncertainty-Aware Probabilistic Travel Time Prediction for On-Demand Ride-Hailing at DiDi. In KDD. 4516–4526

    2023

  20. [28]

    Hongbin Liu, Hao Wu, Weiwei Sun, and Ickjai Lee. 2019. Spatio-Temporal GRU for Trajectory Classification. InICDM. 1228–1233

    2019

  21. [29]

    Yiding Liu, Kaiqi Zhao, Gao Cong, and Zhifeng Bao. 2020. Online anomalous trajectory detection with deep generative sequence modeling. InICDE. 949–960

    2020

  22. [30]

    Ilya Loshchilov and Frank Hutter. 2019. Decoupled Weight Decay Regularization. In ICLR. https://arxiv.org/pdf/1711.05101

    2019 arXiv

  23. [31]

    Jointlycontrastive representation learning on road network and trajectory

    ZhenyuMao,ZiyueLi,DedongLi,LeiBai,andRuiZhao.2022. Jointlycontrastive representation learning on road network and trajectory. InCIKM. 1501–1510

    2022

  24. [32]

    Tomás Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Ef- ficient Estimation of Word Representations in Vector Space. In ICLR. https://arxiv.org/pdf/1301.3781

    2013 arXiv

  25. [33]

    Faststochastic routingundertime-varyinguncertainty

    SimonAagaardPedersen,BinYang,andChristianS.Jensen.2020. Faststochastic routingundertime-varyinguncertainty. TheVLDBJournal 29,4(2020),819–839

    2020

  26. [34]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. InNIPS. 5998–6008

    2017

  27. [35]

    Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph Attention Networks. InICLR. https://arxiv.org/pdf/1710.10903

    2018 arXiv

  28. [36]

    Michail Vlachos, Dimitrios Gunopulos, and George Kollios. 2002. Discovering Similar Multidimensional Trajectories. InICDE. 673–684

    2002

  29. [37]

    XiaodongYan,TengweiSong,YifengJiao,JianshanHe,JiaotuanWang,Ruopeng Li, and Wei Chu. 2023. Spatio-Temporal Hypergraph Learning for Next POI Recommendation. InSIGIR. 403–412

    2023

  30. [38]

    Chengcheng Yang, Lisi Chen, Hao Wang, and Shuo Shang. 2021. Towards Efficient Selection of Activity Trajectories based on Diversity and Coverage. In AAAI. 689–696

    2021

  31. [39]

    Fastmapmatching,analgorithmintegrating hiddenMarkovmodelwithprecomputation

    CanYangandGyözöGidófalvi.2018. Fastmapmatching,analgorithmintegrating hiddenMarkovmodelwithprecomputation. InternationalJournalofGeographical Information Science32, 3 (2018), 547–570

    2018

  32. [40]

    Song Yang, Jiamou Liu, and Kaiqi Zhao. 2022. GETNext: Trajectory Flow Map Enhanced Transformer for Next POI Recommendation. InSIGIR. 1144–1153

    2022

  33. [41]

    Sean Bin Yang, Chenjuan Guo, Jilin Hu, Jian Tang, and Bin Yang. 2021. Unsu- pervised Path Representation Learning with Curriculum Negative Sampling. In IJCAI. 3286–3292

    2021

  34. [42]

    Sean Bin Yang, Chenjuan Guo, Jilin Hu, Bin Yang, Jian Tang, and Christian S. Jensen. 2022. Weakly-supervised Temporal Path Representation Learning with Contrastive Curriculum Learning. InICDE. 2873–2885

    2022

  35. [43]

    Sean Bin Yang, Jilin Hu, Chenjuan Guo, Bin Yang, and Christian S. Jensen. 2023. LightPath: Lightweight and Scalable Path Representation Learning. InKDD. 2999–3010

    2023

  36. [44]

    Di Yao, Haonan Hu, Lun Du, Gao Cong, Shi Han, and Jingping Bi. 2022. Traj- GAT: A Graph-based Long-term Dependency Modeling Approach for Trajectory Similarity Computation. InKDD. 2275–2285

    2022

  37. [45]

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

    2017

  38. [46]

    HanyuanZhang,XinyuZhang,QizeJiang,BaihuaZheng,ZhenbangSun,Weiwei Sun, and Changhu Wang. 2020. Trajectory Similarity Learning with Auxiliary Supervision and Optimal Matching. InIJCAI. 3209–3215

    2020

  39. [47]

    QianruZhang,ZhengWang,ChengLong,ChaoHuang,Siu-MingYiu,YidingLiu, Gao Cong, and Jieming Shi. 2023. Online Anomalous Subtrajectory Detection on Road Networks with Deep Reinforcement Learning. InICDE. 246–258

    2023

  40. [48]

    WheretoGoNext:ASpatio-Temporal Gated Network for Next POI Recommendation.IEEE Transactions on Knowledge and Data Engineering34, 5 (2022), 2512–2524

    Pengpeng Zhao, Anjing Luo, Yanchi Liu, Jiajie Xu, Zhixu Li, Fuzhen Zhuang, VictorS.Sheng,andXiaofangZhou.2022. WheretoGoNext:ASpatio-Temporal Gated Network for Next POI Recommendation.IEEE Transactions on Knowledge and Data Engineering34, 5 (2022), 2512–2524

    2022

  41. [49]

    YushengZhao,XiaoLuo,WeiJu,ChongChen,Xian-ShengHua,andMingZhang

  42. [50]

    Yu Zheng, Xing Xie, and Wei-Ying Ma. 2010. GeoLife: A Collaborative Social NetworkingServiceamongUser,LocationandTrajectory. IEEEDataEngineering Bulletin33, 2 (2010), 32–39

    2010

  43. [51]

    Silin Zhou, Jing Li, Hao Wang, Shuo Shang, and Peng Han. 2023. GRLSTM: Trajectory Similarity Computation with Graph-Based Residual LSTM. InAAAI. 4972–4980

    2023

  44. [52]

    Yu Zhu, Hao Li, Yikang Liao, Beidou Wang, Ziyu Guan, Haifeng Liu, and Deng Cai. 2017. What to Do Next: Modeling User Behaviors by Time-LSTM. InIJCAI. 3602–3608

    2017

  45. [2015]

    InarXiv preprint

    Review and perspective for distance based trajectory clustering. InarXiv preprint. http://arxiv.org/pdf/1508.04904

    arXiv

  46. [2023]

    Dynamic Hypergraph Structure Learning for Traffic Flow Forecasting. In ICDE. 2303–2316

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.